logoPuppy developer news:

4.00alpha1 to 4.00alpha3

left-arrow Older news

arrow-rightLater news

Dingo 4.00alpha3

December 14th, 2007

Puppy Dingo 4.00alpha3 is available. To provide distinction from other releases, it has version number 393, hence the ‘devx’ file is ‘devx_393.sfs’. The live-CD is file is 80.26MB, which is rather “big”, but this build has the full ‘zdrv’ file. Download from here:


The main feature relative to the alpha2 release is the move to the 2.6.24-rc4 kernel, with radical changes such as SMP, kernel-level PCMCIA and the putting to rest of the /dev/hd* drive notation — see announcement about this kernel earlier in this blog. Thus, the main focus with testing this release will be how well does the kernel perform:
1. Boot on old and new hardware?
2. Recognise all drives?
3. Does PCMCIA work?

The usage of the 2.6.24-c4 kernel is only temporary and the final release will have 2.6.24 or final kernel releases.

In /etc/rc.d/PUPSTATE, the variable SATADRIVES now includes any IDE hard drives, as Puppy can no longer tell the difference. I will be doing a global change of that variable name. All internal hard drives and USB drives are now /dev/sd*, all CD/DVD drives are now /dev/sr*. Note, the SATADRIVES variable does not list any USB drives.

tempestuous got me going with a package that provided the proper PCMCIA ‘pcmciautils’ support for the 2.6 kernel, instead of the old 2.4-based system that we have used in all previous puppies. Does it work though? If not, why not?

Alpha3 has the Xorg drivers as well as Xvesa, however there is only a cutdown selection of Xorg X servers — look in /usr/X11R7/lib/xorg/modules/drivers. Your video hardware will most likely work with the ‘vesa’ server if there is no match, but most likely you will want a server for your specific hardware. In that case, grab the full Xorg 7.3 binary package that I have uploaded to ibiblio (in the Dingo alpha2 folder) and get your required driver out of it. If you have already had the ‘vesa’ server running, you might want to delete the /etc/X11/xorg.conf and xorg.conf.xxxx files before rerunning the ‘xorgwizard’. I would like to know if your specific server works okay and in particular does it need any other files. I will include more servers in the next release of Dingo.

There are lots of small notes… The frugal install with save-to-entire-partition is not yet fixed. The kernel has been patched with squashfs and unionfs, no other patches and is verbose when booting, no third party drivers compiled. No PET packages uploaded to ibiblio yet, so the PETget package installer will not yet work. I have put in a couple of the alternative proposed logos, just to see how they look.

As usual for an alpha, do not upgrade any pre-existing pup_save file (use ‘puppy pfix=ram’ at bootup), and it is for our puppy-testers, not for general release.

19 Responses to “Dingo 4.00alpha3”

  1. Sage Says:

    Interesting. GPartEd stalls until a FD (any) is inserted. Cfdisk not working, but sometimes it does this! If, with an IDE drive present, one chooses to install to internal IDE, it says no drive found, but if an internal SATA is selected, even though it’s an IDE PATA, installation proceeds normally. GRUB installation is quirky - had to do it manually first time. The nomenclature will need editing at some stage in all these menus. Failed to start up due to looping at

    Starting X……
    3 lines referring to /usr/X11R7/bin/xwin, lines 311, 312 & 341
    Exited from X….
    (To shutdown PC type…..) message
    We had this issue once on a previous recent release which turned out to be a simple fix? [But not for me!].
    Next - the acid scsi test…..

  2. pogo2008 Says:

    Hi Barry,

    used xorg on 2 computers, vesa on third. Errors on firewall:

    …iptables: Index of deletion too big
    Building firewall iptables: Invalid argument repeated 3 times.

    Also on all 3 computers recognized eth0 & claimed to find alive network & get DHCP

    but unable to reach the internet with seamonkey???

    All 3 worked with Puppy 3.0???



  3. BarryK Says:

    Sage, yes, I should have mentioned that in the release announcement — if you use the Universal Installer, don’t bother with selecting to install to an IDE partition. Well, I already did state this, Puppy now cannot distinguish IDE hard drive from a SATA hard drive. I will be removing the ‘Install to IDE hard drive’ choice from the menu.

  4. Sage Says:

    So far, continued to fail to get the installed version running despite (feeble) attempts to edit or copy stuff from CD - beyond my ability level. Don’t know where to look.

  5. fmlynch Says:

    No problems booting from CD. upgrades from a2. The Prism2_usb driver seems to be missing from ZDRF?

    F M Lynch

  6. Sage Says:

    Cannot get connected. Static IP, everything double/quadruple checked - router ping OK. Neither browser or ibilio respond. All other machines active and functionally connected on same line via router and hubs. Haven’t seen this issue for a long while.

  7. plinej Says:

    Is there a ndiswrapper module in the zdrv? Just want to be able to get online with this alpha. Just asking because I didn’t see one.

  8. cherriepuppy Says:

    Well done barry and the puppy team, burned to cd on main pc (athlon 3000, ati9250 graphics,1GB ram), booted up puppy (pfix=ram), extracted the neomagic driver from the full Xorg 7.3 binary package and saved in /usr/X11R7/lib/xorg/modules/drivers. shut down pc saving session to cd.
    now the good bit….put cd in toshiba portege 7010 laptop and booted up using wake2pup, Xorg wizard asked to confirm lcd resolution, tested ok,ctrl+alt+backspace, done, puppy running fine. First time without having to mess around altering xwin and manually configuring Xorg.conf. this will be great for all puppy users with older toshiba laptops

  9. veronicathecow Says:

    Success! Nice one Barry, with this new Kernel I get all my video modes (Still got the tearing of the screen problem that all linuxes get)
    Sound works fine, Internet fine, everything from a hardware point of view seems great. (Manual Frugall Install)
    Will try fast boot etc now.
    Many many thanks
    Tony on Intel’s D201GLY MB

  10. BarryK Says:

    ‘ndiswrapper’, ‘prism_usb’: these are third party drivers, as stated in release announcement they have not been compiled.

  11. BarryK Says:

    ‘Firefox’ has started a forum thread for feedback on Puppy Dingo alpha3:

  12. ly2101 Says:

    I use it on my computer with mainboard i440mx,cpu CR500,4M video card.
    It has better compatibility with old hardware than previous version.
    Thank you,BarryK!

  13. GeorgR Says:

    I am writing on a PC with AMD690G Chipset and Puppy running.
    This is the first Puppy version booting on AMD690G without any problems and no need for special boot options.
    All drives are recognized.

    I am blinded by the yellow collor, but when looking out of my real window on to the grey clouds, I am really happy about the sun inside Puppy. ;-)))

    ASUS M2A-VM, AMD 690G, ATI SB600, ATI Radeon X1250, Audio: ALC883, LAN: RTL8168/8111

  14. capoverde Says:

    Dingo’s first run today here. For me, too, it’s the first time I can’t get an Internet connection with Puppy - which is always my No.1 test.

    The setup wizard finds my ethernet card and sets it to work as usual - or at least so it seems; but neither SeaMonkey nor NetSurf can load any remote page… Will try harder. Only sure thing for now is that the .iso file isn’t corrupted (md5sum OK).

    Mainboard: ASRock 775i65GV with Celeron 2800, embedded RTL-8139 type ethernet card.

  15. Sage Says:

    Static IP connection, and serial modem connection according to Raman, is broken, capo. Cannot see the point of any further testing or comments until that is fixed! Distro building is a very skilled operation and Barry needs much more help with such core issues.

  16. veronicathecow Says:

    Hi, I have had problems with the firewall stopping Internet access completely. Please see the Main forum for details. Perhaps this is causing the problems capoverde and Sage are experiencing?

  17. Sage Says:

    Sadly, not. I’ve tried everything I can think of, including editing files, wholesale copying of files and folders from previous alphas, betas, 3.01 (although the Greg fix for /etc had some effect). Nothing works - this one is totally broken on all the counts I described and a whole bunch reported by others. This is a major concern, and like I said, Barry needs massive help from those who can.

  18. capoverde Says:

    Thanks Sage for sparing me useless tinkering. Anyway, the problem isn’t just with static IP: my connection is over DHCP.

    The wizard finds and sets up the net card as usual, which makes me think it might be a trivial problem with a wrong device name, such as would happen by trying to access a partition with the former “dev/hda” name instead of “sda”. But probably it’s too simple to be true, and already checked out?

  19. capoverde Says:

    I’m sending this with Dingo! Thank you Tony/Veronicathecow, in my case it *was* the firewall — which I didn’t even remember having activated. Connects OK without; no-go with it.

Modified volume control for Dingo

December 14th, 2007

I would like to thank HairyWill for providing a temporary version of his volume control applet. I announced this applet earlier in this blog, see that for links. The problem was that I thought the latest version, 0.6, uses a bit too much CPU time, so Will kindly added the ‘-bg’ commandline option to his first creation, v0.1, which has much less CPU usage, and this will be adequate for inclusion in alpha3.

MP3info, Pmetatagger

December 14th, 2007

MP3info is a program to edit ID3 tags in MP3 audio files. The package has both console and GTK2 GUI applications but I have only kept the former. Home page:


Plinej has created a GUI application to edit tags in MP3, OGG/VORBIS or FLAAC audio files. This uses commandline utilities, which is why I had to obtain the MP3info package. Jason’s program is called ‘Pmetatagger’ and is version 0.3. Forum page:


3 Responses to “MP3info, Pmetatagger”

  1. Dougal Says:

    I’ve been meaning to mention this in the forum, but keep forgetting: I downloaded the mp3info code and it only supports id3v1 tags.
    It would be better to use the id3info utility (part of the id3lib package) — the output is also friendlier and will make the code simpler.

  2. plinej Says:

    Thanks Dougal, I’ll mess around with id3info soon.

  3. plinej Says:

    I posted version 0.5 with id3lib utils support for mp3 id3v2 tags. I also fixed some bugs so I recommend checking out this new version.

Script languages in Dingo

December 14th, 2007

I have agonised over this. The Dingo project was started as a total rethink of Puppy, built with only what is absolutely essential. I considered a Qt-only Puppy, but soon realised that there are far more GTK2-based applications, so chose a GTK2-only system. I threw out Tcl and PuppyBasic, which left only the Bash and Perl scripting languages. The GUI libraries Tk, Gnocl and GTK1 were removed, but libXaw is still there, for simple Xlib GUIs.

My thinking was that I would create the leanest Puppy possible, and in future others could bring out fatter puppies. I have even built up Xorg file-by-file, to create the smallest-ever Xorg package, without OpenGL, DRI — currently also with cutdown selection of input and video drivers. I am still planning to take some time off, probably in the first half of 2008, with continued but reduced involvement while I travel, and others can be creative and bring out fatter puppies.

A problem before was that I put stuff into Puppy just to please someone. This new lean Dingo will not please everyone, however I do see that puplets can be created with those extra things included, to satisfy those with particular interests. One of the main issues has been the scripting languages used in Puppy, and the murgaLua and Gtkbasic projects are so good that I feel really bad leaving them out — however, whichever way I look at it, I can’t justify putting them into the base Dingo system. Dingo has Bash and Perl, and basic GUI-builder utilities, and these have proven to be quite satisfactory.

So, that’s how I am positioning Dingo. When the final version gets released, I intend to release a ’standard’ Dingo which will have a selection of applications for normal use, like SeaMonkey, Abiword, etc. Also there will be a ‘barebones’ which will be just enough to get X running and get onto the Internet, and this can be the core for creating fatter custom puppies — either by remastering or using Unleashed.

Of course I intend to create PET packages of the latest murgaLua and Gtkbasic, so any PET application created with these can specify it as a dependency and it will also get automatically installed.

When Dingo is out and reasonably bug-free, I’ll take my break.

7 Responses to “Script languages in Dingo”

  1. MarkUlrich Says:

    I have no problem with your decision concerning GtkBasic :)
    Instead I could imagine that GtkBasic will become a part of EzPup, the addon that will add additional themes and wizards.
    My Configcenter makes progress, and it will be the base for a icewm configuration-tool.

    More important (very important) it would be, that you replace the english strings in your wizards with variables, that are definded in external .mo files, or at least in the beginning of your scripts.

    It is really a hell to modify them completely from new when you upgrade them, as it currently just seems to be the case with xorgwizard.
    I had started to localize it:
    So for the next Muppy, I have to start from scratch, what is very time-consuming.

    Apart from that, it is very interesting to see Dingo evolving :-)

    Greets, Mark

  2. BarryK Says:

    Mark, I was editing my post while you posted a comment, so it has been modified slightly. I’ll create PET packages to make it easy to customise Puppy, including a Gtkbasic package.

    Regarding customisation of xorgwizard, that has some recent bugfixes, so I’ll be happy to cooperate with extracting the strings based on the puppy4 released version of xorgwizard.

  3. MarkUlrich Says:

    fine, thanks, if you could do that, I could try to backport it for Puppy3.


  4. Vettephil Says:

    “Also there will be a ‘barebones’ which will be just enough to get X running and get onto the Internet”

    This excites me and I think is the perfect way to go for a FOSS OS, allowing users to completely customize and setup their environment to their liking and only including what they see fit. Ever try to uninstall IE on Windows? ;-) Keep up the great work!

  5. GreatnessGuru Says:

    “Dingo has Bash and Perl, …”
    What!? No Postscript?!

    I can’t get Dingo 4.0.0.Alpha.3 booted (see forum) to test for myself, but if a stock Dingo, ’standard’ or ‘barebones’, does include Postscript (Ghostscript…) please remember to mention that in your docs and “marketing campaign”, too. Ok?

    Thank you,
    Eddie Maddox
    Postscript fan

  6. rarsa Says:

    It is really good to have a set direction. With that in mind is easier to contribute whenever we can.

    Some time ago I thought that the go forward was tcl, so I took the time to learn it. Now that I see that it is really Perl, I’ll take the time to learn it.

    Puppy is a really nice project and I don’t mind it being a moving target.

  7. BarryK Says:

    Yes, of course, Puppy supports Postscript. It’s a nice language in its own right, just tends to get forgotten.

    Perl is there not because I want it, but because it has to be. CUPS and ndiswrapper need it. The ’standard’ Puppy has a cutdown Perl, just enough to run these apps. We won’t be putting any Perl addons, like GTK plugin — anyone who codes in Perl can use the Xdialog and gtkdialog GUI-builder apps, just like Bash coders do.

Pburn, Pcdripper, Pfind, Pnethood

December 14th, 2007

All of these are updated. These guys all have an enthusiasm for making Puppy even more special. What really stands out with these apps I think is the very small size, as maximum usage is made of what is already in Puppy.

Pburn v0.4, a CD/DVD burner created by zigbert.
Pcdripper v3.0, a CD ripper created by plinej (previously pbcdripper).
Pfind v3.3, file and directory find created by zigbert.
Pnethood v0.5, to access Samba shares created by HairyWill.

Pburn and Pnethood need special mention, as these have been created to fill a void where we cannot find anything suitable. Yes, there are CD/DVD burner apps out there, but they either have enormous dependencies (like KDE) or are buggy and lapsed or partially-lapsed projects. For accessing shared network resources such as hard drives, especially with Windows computers, we have had limited choices, like LinNeighborhood which is a GTK1 application, and HairyWill has taken on the challenge of a replacement, which is Pnethood. Both of these projects have been very successful, and I can’t show enough appreciation to these guys.

Pburn forum page:

Pnethood forum page:

Pctorrent updated

December 13th, 2007

Plinej (Jason) is a consistent long-time contributor to the Puppy project and deserves recognition. As well as developing applications, he makes the time to help on the forum.

Dingo alpha2 has ‘Pupctorrent’ and ‘Pupcreatetorrent’, which are GUIs for ‘ctorrent’. That was version 0.9, not so long ago, but Jason has been busy improving it and it is now up to version 1.3, as well as a name change to ‘Pctorrent’ and ‘Pcreatetorrent’. He also created a mime-association for ROX-Filer. This new version will be in alpha3.

Read more about Pctorrent in the forum:


One Response to “Pctorrent updated”

  1. plinej Says:

    Barry, I also updated PBcdripper (renamed to Pcdripper) with lots of improvements.


    I made a simple id3 tagger for ogg, flac, & mp3 as well.


Freememapplet bug, want predecessor

December 12th, 2007

I am not using ‘freememapplet’ in Dingo, as after recompiling it with GTK2 (it’s a GTK1 application), I found that it uses a lot of CPU time, and ‘top’ shows it right at the top of the list. That’s version 1.2.2-1, and the source can be found at puptrix.org.

Instead, I’m displaying free memory at the top of the screen, with transparent background so it doesn’t interfere much with icons and windows. I already mentioned earlier in this blog that the text display changes to more detailed information as the free memory gets low. That fine to have the urgent message at top of screen, but I would still prefer to have the regular free memory display in the tray.

So, I got the source for freememapplet v1.0, compiled it with GTK2 and this behaves much better, ‘top’ shows it sitting just above blinky and below rox.

But, I would like to compare with the original free-memory applet that I developed. This was called ‘wmpupmem’ and I developed it by modifying a WindowMaker applet called ‘wmapmmon’, that uses only Xlib. Lior Tubi rewrote my program for GTK1 and freememapplet 1.0 was born.

I’ve been hunting through my archived DVDs, can’t find the source for ‘wmpupmen’, googling can’t find it either. Does anyone by any chance have the source?

5 Responses to “Freememapplet bug, want predecessor”

  1. pakt Says:

    Forum member ‘amigo’ just reported “I have a site with a few hundred GTK-1.2 apps”.

    I found the following in one of the sub-folders


    Could it be what you’re looking for?

  2. krumpli Says:

    could this be program:

  3. BarryK Says:

    No, guys, I want the source for my application, ‘wmpupmem’, as stated in the first post. I’ll hunt further through my archived DVDs, I must have it somewhere.

  4. krumpli Says:

    Off topic, but in an earlier post you discussed media players. Below is link to a 6 part review of Linux media players.


  5. MarkUlrich Says:

    I added trayicon support to GtkBasic.
    I cannot serve you with a freememapplet unfortunately, as I am too busy at moment.
    Maybe next week.


Xorg Wizard bugfix

December 12th, 2007

A couple of things fixed/changed.

When you test the video mode, a dialog box shows the resolution and vertical and horizontal frequencies, but sometimes the last two had an incredible number of digits to the right of the decimal point, for example ‘60.123456′. I’ve rounded that off to two digits.

Rerwin and Dougal have been helping people with laptops for which the Xorg Wizard either hangs or offers inadequate resolutions. They have worked out a fix, see this forum thread:


I’ve implemented an equivalent fix, not the one I posted near the end of that thread as rerwin tested that and found it wanting. So, I made some more changes, and it will be in alpha3 for testing.

One Response to “Xorg Wizard bugfix”

  1. raffy Says:

    While in the subject of X and desktop, gOS has earned praise for its very user-friendly task bar:
    “The gOS iBar is located at the bottom of the screen and features quick launchers for many popular Web sites and services…”

    I suddenly recalled that this kind of desktop was already implemented by lvds in the Forum, in eXpand Barbie: http://www.murga-linux.com/puppy/viewtopic.php?t=17810 . Forum discussion is here: http://murga-linux.com/puppy/viewtopic.php?t=24422

Fastboot now faster

December 11th, 2007

This is very interesting. I reported about fastboot earlier, see this blog post:


With fastboot turned on, I was getting a boot time of 19 seconds, with Puppy installed on a SATA hd, full install, timed from making the selection in the GRUB menu to desktop fully loaded.

Well, exactly the same installation, but I compiled the new 2.6.24-rc4 kernel and when I rebooted I noticed it seemed to boot remarkably fast. In fact, everything seems faster. Anyway, I timed the boot, it is now 14 seconds. This includes DHCPCD network connection, well almost, as the desktop loads and there may be a second or two before Blinky shows an active network — which is dependent on how fast the DHCP server responds.

13 Responses to “Fastboot now faster”

  1. veronicathecow Says:

    Great news Barry. See, we’ll get Puppy booting and ready before the power button is pressed eventually. Lets see, Capacitance sensor fitted to the computer chair….. Linux BIOS, Infinite Improbability Drive…

  2. veronicathecow Says:

    Seriously though, when this is stable, what a great press release this would make! Get a CE Edition with additional stuff, O.O. Thunderbird, Firefox etc. I would be happy to do some press releases in UK. Perhaps we could compare Puppy boot up and running speeds on a machine that Microsoft consider inadequate for Vista with Vista on a current consumer machine including costs!

  3. raffy Says:

    Good news. :)

    Xandros in the eee is still the time to beat: 15 seconds. One or two Puppy Linux enthusiasts will receive their own unit soon and begin tests.

    Hey, Barry, why not exchange the other Classmate PC with an eee PC? Surely, the Intel manager who contacted you can arrange that.

  4. raffy Says:

    I mean, how do we know what “15 seconds” mean until someone from the Puppy community has done the tests…

  5. veronicathecow Says:

    Hi Barry, just came across this in a mag. It’s not stable yet (See recent notes on web page) but it sounds very useful for Puppy flash system.


  6. cb88 Says:

    barry I doubt you have seen this so here it is…. the eee pc running compiz!


    most people are gonna want the 3D effects…. maybe the components of wNOP could be made into an sfs…?

  7. Dougal Says:

    Raffi, I assume the idea of having two Classmates is so he can get mesh networking working between them.

  8. gnomen Says:

    What about scrapping the whole init-system and go for eINIT http://www.einit.org/ It would be a major undertaking, but well worth it in the long run. I am currently trying it out on my personal desktop (not puppy). It is quite stable, and it really makes a huge difference. It would not surprise me if you could half the boot-time on our beloved puppy

  9. veronicathecow Says:

    Hi Gnomen, looks interesting but there seems little activity on the site. (Last forum post 8 weeks 4 days ago by jyujin and few other posts) so will there be support? There are some other ideas like upstart
    or search the main forum for turbo puppy for somer other thoughts.

  10. bostonvaulter Says:


    I just received an eeepc recently and mine boots in about 20 seconds not including wireless.

    I’ve installed puppy to it, but not very well and haven’t enabled fast boot.

  11. veronicathecow Says:

    Apparently this is supposed to be fast.
    Using the following optimizations

    0.) Performance patches from Con Kolivas, i686 2.6.7 kernel, reiserfs
    1.) All original sources, minimal patches.
    2.) Compiled with i686 against latest gcc (Did you try compiling for 686?)
    3.) Stripping
    4.) Prelinking
    5.) Latest gcc and glibc and other sources
    6.) Keep everything only dependent to what it really needs not what the ./configure happens to find.
    7.) Hdparm on install

    I believe texas flood boot operates by setting up CD to do a continuous read from continuous tracks, can a Puppy HD install do the same?

  12. penk Says:


    I did it in 10 sec on Ubuntu Linux,
    (The X starts from 7 sec.)
    with modified InitNG and Yaird.

    Here’s the boot chart:


  13. veronicathecow Says:

    Hi Penk, that’s amazing. How did you do it and can any of that be applied to Puppy?
    I notice it’s only a 1.86 Ghz celeron so that even more amazing!

/dev/cdrom, /dev/dvd symlinks fixed

December 11th, 2007

Running my new 2.6.24-rc4 kernel without /dev/hd* devices, I found that the above symlinks were not getting set.

I examined /etc/rc.d/rc.local0 and found the code for setting those symlinks assumes /dev/hd* drives. So, I have generalised the code to recognise any CD/DVD drives. This code improvement is backwards compatible with older puppies.

Note, I’ve compiled Xfmedia, the XFCE media player, but haven’t tested it yet. My morning got taken up with the Great Gxine Saga, now I’m onto other stuff.

In answer to Dougal about reporting the Gxine bugs, I haven’t done so. Most likely the crashing bug can be fixed, but the lack of full-screen support has been an on-going saga — I did report this problem — ages ago, maybe 1.5 - 2 years — but it is still an issue. I seem to recall that after a version upgrade, full-screen stopped working, after a few more versions it was working again, but now it isn’t again. So, I’m not in any hurry to report it again. Also, there was a peculiar flickering that came in at some stage, the image suddenly has a kind of periodic flicker across it, but I never decided whether that was due to upgrading JWM or Gxine. But it’s interesting that I’m not getting it with Gxine 0.4.9 — but then maybe there were some particular circumstances that made the image seem flicker-free, and others will have to test and compare after alpha3 is released — at this stage I’m intending to put Gxine 0.4.9 into Dingo alpha3.

2 Responses to “/dev/cdrom, /dev/dvd symlinks fixed”

  1. BarryK Says:

    Heh, heh, I should keep a record of all the packages in which I have been forced to use an older version. Let’s see…

    pcitools v2.1.11
    bash v3.0.16
    cups v1.1.23
    gxine 0.4.9

    At one stage I was forced to use an older ‘crecord’ from cdrtools, but I think that is fixed now that I’ve moved over to the cdrkit package.

    In the case of ‘lspci’, the developer changed the output format which broke some of my scripts, and I have stubbornly refused to fix my scripts, instead keeping the older lspci.

    cups, well that is documented recently in this blog.

    bash, the 3.1.x series breaks some scripts. Fixable, but once again I can’t be bothered as I’m quite happy with v3.0.16.

  2. kirk Says:

    To get Gxine 0.5.11 to full screen, first maximize Gxine’s window and then hit ctrl-f. If you hit ctrl-f without maximizing the window, it just flickers.

    I’ve been passing “combined_mode=libata” when booting to get DMA working with my DVD. After doing this my DVD is /dev/sr0. I’ve been just symlinking /dev/sr0 to /dev/dvd and Gxine is happy with that. At least in 2.17 and 3.01. Sounds like dingo will be doing all that for me. Thanks.

The Great Gxine Saga

December 11th, 2007

This has taken me the entire morning, but I now have a version of Gxine that works reasonably well.

Gxine versions are really patchy, a later version does not mean a better version. I know this from previous experience.

The latest version, 0.5.11, crashes when File –> Preferences is chosen. Also, it doesn’t do full-screen — there’s just a weird flicker and it stays the same size. So, I went back, compiling old versions and comparing…

Same problems as 0.5.11.

v0.5.8, 0.5.9
Preferences works, but still no full-screen.

v0.5.7, 0.5.6
Crashes at startup.

Works ok, except doesn’t recognise my /dev/cdrom link to /dev/sr0 as being a CD drive. DVD plays ok though. No full-screen.

This is the last of the 0.4.x series, and this one works! It has the same problem of not recognising the CD drive, and the control-panel in the Gxine window is a bit reluctant to render  — sometimes have to jiggle the window to get it  (volume control,  start/stop buttons, etc.) to draw. It does full-screen very nicely.  In fact, there’s something else,  I’m not sure about this but I think recent verions of Gxine have a bit of a flicker when playing video, which is not there in this version.

The Gxine site does not have this, I have it as it was used a long time ago in Puppy. Has the non-recognition of the CD problem, full-screen works momentarily then crashes.

Well, well, I’m now playing a CD with v0.4.9. When I chose the CD, an error window came up with “Connection refused: Operation now in progress” but then it went ahead and played the CD. I can choose tracks too. Probably something to do with it being  ’sr’ device rather than ‘hd’ (it’s an IDE drive but I’m using the new 2.6.24-rc4 kernel).  It should be easy enough to hack the source to get rid of that error message.

I tested the web browser plugin from v0.4.9, it works.

I tested various media files. Actually,  Real Video support is not all that bad — I have three files locally for testing and MPlayer plays all three, Gxine plays two of them — both sound and video.

One Response to “The Great Gxine Saga”

crafty Says:

This might be O.T.. - but is it possible to keep the ‘/proc/ide’ folder and somehow symlink to the corresponding drive points in the new ‘proc/sr* (or whatever it be now called)’ folder - at least till Puppy with this new kernel gets fully “burnt-in” tested - maybe by v4.1 - v4.2..

OR if ’symlinks’ does not work - then maybe a script that can do the same..?

Just a thought..


2.6.24-rc4 kernel

December 10th, 2007

Running a full hard-drive installation of Dingo, I have compiled the 2.6.24-rc4 kernel. With radical changes… the entire IDE drive support, which gave us our /dev/hd* drive names, is now gone, by setting CONFIG_IDE=n. Instead, all of the libata PATA drivers are enabled — all PATA and SATA drivers are selected built-in. This means that all hard drives will now be /dev/sd* and all CD/DVD drives will be /dev/sr*.

What else… yeah, this one will be welcomed by some, SMP (symmetric multi-processor) support is now enabled. I also set CONFIG_PCMCIA_LOAD_CIS=y to hopefully get PCMCIA working. Lots of exciting new wireless support — which is one of the reasons I wanted to go for this very latest kernel.

After compiling and installing the new kernel and modules, I crossed my fingers and rebooted. The PC has one IDE and one SATA hard-drive, one IDE DVD drive. Yep, booted up fine, except the DVD drive had gone awol. Ah, the ’sr_mod’ module needs to be loaded, did that and there was the DVD drive, it played a DVD ok also.

It seems pretty obvious that it would be best if ’sr_mod’ is built into the kernel, so I’m recompiling right now with that change.

Some scripts may be broken or partially broken, as the entire /proc/ide directory is now gone.

My plan is to release alpha3 with this 2.6.24-rc4, and anyone who has a special wish for a change in the kernel config may submit it as I’ll do one more compile of the 2.6.24-final kernel before releasing Dingo-beta.

11 Responses to “2.6.24-rc4 kernel”

  1. Sage Says:

    Need I ask?! Did bob9’s package ever arrive?
    Other distros, eg Sidux, with the new nomenclature all manage my old dinosaurs. But, then, DSL always did, even with the 2.4 kernels…..

  2. veronicathecow Says:

    Thanks Barry this will be great to try as I am suffering Puppy withdrawal symptoms.
    Please excuse my ignorance but this seemed a possible idea for keeping Puppy RAM footprint small but I would imagine it would require a number of changes to Puppy. For instance are all file systems needed or could some be loaded at the time access is requested, fat32, NTFS, Reiser (I assume most people are EX2 or EXT3. Perhaps this could be part of fastboot?


    “Enable the Loadable kernel modules support! With this option you can load/unload the device drivers dynamically on running linux system on the fly. See the Modules chapter at Section 3 . “

  3. BarryK Says:

    sage, no, I never received a scsi hardware package.

    Right now I’m running Puppy from a USB Flash drive with the new 2.6.24-rc4 kernel. It’s working perfectly, sound, everything.

    It was easy to fix the ‘probepart’, ‘probedisk’ and ‘init’ scripts to work with the new kernel (and retain backwards compatibility).

    Actually, I don’t expect any problems with other scripts. The ‘SATADRIVES’ variable in /etc/rc.d/PUPSTATE now is not just SATA drives but a list of all internal drives, SATA and IDE. The Universal Installer should be easy to change to handle this unified system.

    One thing I don’t know about is IDE ZIP drives, where before Puppy loaded the ‘ide-floppy.ko’ module — there must be something equivalent. This will need to be fixed in the ‘init’ script in the initial ramdisk.

  4. Sage Says:

    I have emailed bob9. Jesse’s details in your PM box. Surprised you and he hadn’t been able to establish a rapport, although I think his legit. employ keeps him busy 25hrs/dy?
    Delete this if you wish.

  5. Dougal Says:

    I’d be very careful about using the latest kernel before the stable version is released and a few weeks go by: they made a truckload of changes this time around and the patches between rc’s seem to be pretty big, too.

  6. GreatnessGuru Says:

    Dougal Says:
    “I’d be very careful about using the latest kernel …”

    Well, I’ld be very careful, too, about using the latest Puppy…

    That is, we Initial Adopters know things will need fixing and tweaking here and there, but the Puppy and kernel developers have pretty much done their homework before they release their stuff. I cannot say that for all such projects.

    Good call, Barry. I look forward to giving Puppy 4.0 a spin. Perhaps a maintainance release a week or two later with a 2.6.24.z kernel and Puppy touchups would be nice, while you work on Puppy 4.1.

    Oh, if anyone wants to know, I have already decided that Puppy 4.x will be my “upgrade path” from this Mac OSX 10.3.9 system I’m currently using. I have a few legacy Pentium boxes ready to go. The later Puppy 2.x, 3.x, have done well on these. Thanks again, Barry.

    Perhaps a PPC arch Puppy 4 someday? What else am I going to do with my G3? Run Debian?

    Oh #2, I have a friend who is running Puppy 3.x on his Mac in a virtual environment of some kind. He likes it. Good performance.

    Thanks all,
    Eddie Maddox

  7. nic2109 Says:

    Puppy 4 on on a PPC platform………….. now that’s INTERESTING.

    I too know of a G3 iMac that’s not up to MacOS X and is just gathering dust.

    The time is ripe for World Domination. Woof woof!

  8. Headfound Says:

    Could we have ieee/1394 firewire support? I had to use Nathan F’s old unleashed package to get anything to work in 3.

  9. cb88 Says:

    barry i have been looking at directfb alot….

    they have ported GTK to it… I think you see where that is going :-)

    could you see if it will build on puppy 4.00? I am trying to build it in 2.15 with not much sucess….

    I am getting a whole bunch of error when trying to compile matrox gfx drivers…trying disabling gfx in ./configure ..OK compiles correctly now.

    as i understand it multitasking with directfb requires a special kernel module “fusion”….

    I am not really sure what to do to start it up other than that it might need to be passed as a kernel boot paremeter….

    anyway though you might get around to trying out directfb I know it would be a serious undertaking but worth the effort if it really saves as much ram as it claims to…. now i must study for physics test :-)

  10. cb88 Says:

    I forgot one thing… the xvesa man page suggests compiling against ulibc or dietlibc DigitalCrypto mentioned that apps compiled against them are compatable with Glibc… so that may be a way to lower requirements as well

  11. BarryK Says:

    Dougal, yes, I would prefer to wait until v2.6.24.1 before bringing out the final Dingo.

Gxine …grrr

December 10th, 2007

As we have issues with MPlayer in Dingo, I thought that I would give Xine/Gxine a go.

MPlayer is working quite well, except for the disappearing windows with JWM — and I have had no response from the author Joe about this bug, despite two emails (one of them to his sourceforge email address, the other to the JWM mail-list). I guess that Joe is not into working on JWM right now, which is fair enough, considering that it is a free product. I’m quite frustrated, as there are other applications with bugs and I have emailed the authors, no response from anyone.

MPlayer itself has peculiar limitations. For example, the GUI does not have any facility for playing a CD or DVD, only for opening files. A right-click on either window does bring up a more complete menu with option to play a DVD, but no audio CD. However, none of the skins have a “play DVD” button — it seems to be a functionality missing from the skin interface. Also, the “File open” dialog in some skins does not work — I presume that’s due to not being updated for GTK 2.12.

I compiled Xine-lib version 1.1.8 and Gxine 0.5.11 for Dingo. Out of the box, Xine is hopeless with Real Media, files that MPlayer handled without any extra codecs, Gxine just would not play them, not even any error message. Then I selected “File -> Preferences…” from the menu, and Gxine gave a segmentation fault …that’s why the “grrr”. Yep, always gives a segmentation fault, oh well, so much for Gxine.

So, I’m back on MPlayer, it’s a case of choosing the product with less bugs, or non-fatal bugs anyway.

I have put in XfreeCD, a little GTK2 audio CD player, except that is also a problem as it won’t work on all systems. Unlike Xine and MPlayer, XfreeCD requires an audio cable from the CD drive to the audio card, which my laptop doesn’t seem to have. Sigh. Anyone know of a nice little CD player that does direct digital reading from the CD?

10 Responses to “Gxine …grrr”

  1. tempestuous Says:

    Yes, MPlayer is very gui-unfriendly. On the MPlayer developer forum some members even want to eliminate the gui!
    Part of the problem is that MPlayer is very sophisticated, and many options require a long command string. Take, for example, the command for Puppy forum member eden6000’s Digital TV tuner adapter, when he wants to tune analogue TV station 11:

    mplayer tv://11 -tv driver=v4l2:alsa=yes:audiorate=48000:amode=1:adevice=hw.1,0:volume=80:immediatemode=0

    Also, DVD navigation is a recently added feature for MPlayer which requires this command:

    mplayer dvdnav://

    (the libdvdnav/libdvdread libraries from MPlayer need to be installed first)

    It might be worth creating a script with this command, which is launched from the JWM menu.

  2. grumpywolfe Says:

    Hi Barry

    I used xine and one time and liked the interface. I know that you have used JWM on puppy have you given any thought to Xfre. I used it with 3.1 and it was pretty good.

  3. MrPradox Says:

    Hi Barry,

    have you considered the VLC media player? It’s very complete, robust and common even among Windows users…

  4. fwiffo Says:

    What about vlc?

  5. BarryK Says:

    I have just looked at the VLC “Features” page, and it lacks RealMedia support, so is no better than Xine and worse than MPlayer. Also, it needs wxWidgets or Qt4 which is a substantial size overhead.

    Xfmedia, the media player for Xfce, is a possibility. It does not require all of Xfce, just some libraries from Xfce, which does not add much to the size. It uses Xine, so once again has the limitations of Xine. But then, considering that Gxine crashes, Xfmedia is a definite improvement! I might consider putting Xfmedia in Dingo alpha3.

  6. Dougal Says:

    First, I hope you run Gxine in gdb and sent a bug report…

    The reason Mplayer doesn’t have the “play cd” option is that it’s a movie player… it’s supposed to be complemented with something like xmms.

    Xfmedia is a xine frontend, so you’ll still not have RealMedia support…

    What about Xine-ui, by the way? You can find a decent skin for it and then it’s not **that** awful (assuming it works)…

    A few months ago I tried searching for a replacement for XfreeCD and found nothing.

  7. cb88 Says:

    The latest Xmms would make a good adition and john biles recently posted a plugin for xmms that improves CD playback…. and don’t forget Xmms resource usage will always be lower than gxine… how to make Xmms support .wma I am not sure but it is possible…..

  8. erik496 Says:

    xmms wma-plugins can be found in


    it works for me in xmms-1.2.9 and puppy 2.16

  9. tempestuous Says:

    How to make XMMS support .wma?
    Easy, compile the XMMS-WMA plugin from http://mcmcc.bat.ru/xmms-wma/

    Regarding VLC: it’s overrated … perhaps I said that on the forum once, twice, or ten times already?

  10. tempestuous Says:

    Regarding Xine and its variants, xine-ui seems to be more reliable than gxine with fullscreen and various other windowing issues, but xine-ui does not have a Mozilla plugin.

Working with a small screen

December 9th, 2007

I have made an interesting discovery.

I’m testing the Xorg Intel driver on the Classmate PC laptop, and I can get it to run in the native 800×480 resolution of the LCD screen, but I can’t get a virtual screen with panning. I tried all sorts of things in ‘xorg.conf’, to no avail …but I do vaguelly recall some Xorg drivers have a problem with virtual screen and panning …then I came across a comment somewhere that the Xorg developers no longer support panning.

After considerable frustration, I turned to JWM and the pager mechanism. The main problem is the 480 pixels vertically — I really would like to have a virtual screen of 800×600, where the screen scrolls if the mouse pointer is moved to top or bottom of screen. However, failing that, a pager in the tray that is organised as 2×2 would be good. But, it seems that JWM only supports horizontal paging. Bummer.

I then studied the JWM changelog and found a new feature introduced with v2.0 — the windows inside the pager-windows (in the tray) can be dragged around with the right button. This enables quick access to windows that stick off the bottom and side of the screen.

But, there’s another advantage — JWM has a bug in that sometimes windows draw with the titlebar off the top of the screen, so cannot be dragged down by left-button-press-and drag in the normal way. However, right-button-press-and-drag inside the pager works nicely.

Um, well, “nicely” is not the right word. It’s a matter of getting by when there’s nothing better, in fact nothing else at all. It’s actually a crude mechanism, as a tiny movement of the mouse and the window makes a big jump. If I could drag the window up or down smoothly with the scrollwheel would be nice.

Note, I think the Asus eee has same or similar video chip? My  Classmate has a:

Intel(r)915GM/910ML/915MS Graphics Controller

5 Responses to “Working with a small screen”

  1. Pizzasgood Says:

    You can usually also move things in Linux by holding [Alt] and dragging anywhere within their window. I haven’t used JWM in a while, so I’m not positive that it supports that, but I think it does.

    That’s not an obvious method though.

  2. Dougal Says:

    Any idea when alpha3 might be ready?

  3. BarryK Says:

    Pizzasgood, yes! thanks for that.

    Dougal, alpha3, I’m not sure. I want to recompile the kernel and turn on pcmcia plus maybe other changes to the kernel. Maybe another week from now. Ah, no, looking at my schedule, I really need to upload alpha3 by this Friday when I’ll have broaband2 access — well, I’ll see if can meet that deadline. That’s the 14th.

Database GUI for Puppy

December 9th, 2007

Guys, we finally have something. Puppy has two databases, the Berkely database and Sqlite3. What we don’t have is a GUI tool for creating and accessing a database. We have discussed options in this forum thread:


I mentioned a package, named ‘SQLiteDBMS, that is quite small and uses a web-browser interface, however I could not get it to work in Puppy. That was back around mid-2007. Yesterday, forum member ‘lgo AtM’ has reported that he has worked on the source and got it to run in Puppy:


We need testers! Are you interested in a database application for Puppy? Then please help lgo AtM to find and fix bugs.

SQLiteDBMS home: http://sqlitedbms.sourceforge.net/index.htm

5 Responses to “Database GUI for Puppy”

  1. nic2109 Says:

    I’ve not tried SQLiteDBMS on its own, but I have used XAMPP in Puppy and SQLite is part of that. In that guise SQLite worked for me, though I only used really simple SQL. My real use for it was to drive a Wordpress Blog hosted locally.

    The full XAMPP download is 52MB so is out of scope for a base Puppy. Would be good for a CE though!

  2. WhoDo Says:

    I’ve tried to compile SQLiteDBMS on my Puppy 3.01 system, but I keep getting a terminal compile error telling me:

    “configure: error: sqlite3 library is required by SQLiteDBMS. use –with-sqlite3=DIR.”

    Of course I have entered the appropriate directory /usr/bin, and I’ve checked at CLI that sqlite3 runs from that binary, but still no joy.

  3. MarkUlrich Says:

    please try:


  4. WhoDo Says:

    “configure: error: sqlite3 library is required by SQLiteDBMS. use –with-sqlite3=DIR.
    # ./configure –with-sqlite3=/usr/lib/sqlite3″

    Sorry, Mark. No joy, as can be seen above.

  5. MarkUlrich Says:

    ok, you’re right.
    I now did this:
    extracted (found no source for Puppys older version):
    ./configure –prefix=/usr/local/Sqlite-3.5.4 –disable-tcl
    make install

    strip ‘ed the files in bin and lib folders.

    now the dbms
    ./configure –prefix=/usr/local/SqliteDBMS-0.51 –with-sqlite3=/usr/local/Sqlite-3.5.4
    make install

    strip ‘ed the files in bin and lib folders.

    renamed /usr/local/SqliteDBMS-0.51/conf/sqlitedbms.conf.sample to sqlitedbms.conf

    cd /usr/local/SqliteDBMS-0.51/bin
    # ./sqlitedbms
    ** Catalog files not found. Insert admin id and password for initialize catalog.
    Administrator ID:

    Now I’m too lazy to continue and read the faq ;)
    However, maybe it now will work with the old sqlite, or you had to create an additional dotpup for the new one.

    My results so far, extract to /usr/local/

    You might have to set in a console before you continue:
    export PATH=/usr/local/Sqlite-3.5.4/bin:$PATH
    export LD_LIBRARY_PATH=/usr/local/Sqlite-3.5.4/lib:$LD_LIBRARY_PATH

    export PATH=/usr/local/SqliteDBMS-0.51/bin:$PATH
    export LD_LIBRARY_PATH=/usr/local/SqliteDBMS-0.51/lib:$LD_LIBRARY_PATH


Agave Americana in flower

December 9th, 2007

Scroll down through my blog to find the first snapshot, when this enormous stalk suddenly shot up. Now it is in flower.


2 Responses to “Agave Americana in flower”

  1. hairywill Says:

    wow, big asparagus

    Barry, is there any chance you can leave the cifs module in the next zdrv. I’d like to experiment with using cifs in pnethood. If this works smbmount will be redundant and the cifs module can be stored in the zdrv rather pup_xxx.

    cheers, Will

  2. BarryK Says:

    hairywill, ok, done. It will be in alpha3, due out soon.

Desktop icon switcher for Dingo

December 6th, 2007

Well, it can be adapted for other puppies also.

The icons that appear on the desktop are all at /usr/local/lib/X11/pixmaps, so to are the top-level menu icons. The former are 48×48 pixels, the latter 24×24 pixels. What is new in Dingo is I have created /usr/local/lib/X11/themes, which has subdirectories with different sets of icons. Currently I have ‘Outline-svg’ and ‘Stardust’ subdirectories. The former has a set of SVG icons, the latter a set of PNG icons created by zigbert.

What my icon-switcher does is copy the icons from the chosen theme across to /usr/local/lib/X11/pixmaps. It’s a bit more sophisticated than that though. If the source-directory has any SVG icons, they are copied to the ‘pixmaps’ destination directory as 48×48 and 24×24 PNG icons, named appropriately — for example, ‘themes/Outline-svg/www.svg’ would cause ‘pixmaps/www48.png’ and ‘pixmaps/www24.png’ to be created. If the source-directory has a 48×48 PNG image, then it gets copied but also a 24×24 PNG image created — for example ‘themes/Stardust/www48.png’ will get copied to ‘pixmaps/www48.png’ but also scaled down and ‘pixmaps/www24.png’ created.

If anyone wants to add a new set of icons, just create a new subdirectory under ‘/usr/local/lib/X11/themes’ and the script will automatically recognise it.

The script is /usr/sbin/icon_switcher and is written in Bash with a gtkdialog GUI. Here’s a snapshot:


The images shown are just samples taken out of each theme.

11 Responses to “Desktop icon switcher for Dingo”

  1. lobster Says:

    We have two other sets of superb icons that I hope there will be room for. The colourful classic Puppy set and the transparent icons in 2.15CE. Looks as if the SVG icons are evolving too. :)

  2. BarryK Says:

    Lobster, reload your web page, I just fixed the button-widths and uploaded a new snapshot.
    Well, the SVG outline icons still need more work.
    Yes, the original colorful icon set is still very popular, so maybe should be included. I don’t remember what the 2.15CE icons look like…

  3. lobster Says:

    The first thing the majority of people do is change the look and feel. If this is easy and pleasing, then they are easily pleased

    Here are the icons I rather liked from 2.15 beta — they’re called glass

    Here they are evolved in Ezpup

    Hopefully Warren still has a recommended set . . .

    What intrigues me about using SVG is the possibilities. Icons can be resized from one set. There is an animation potential . . . m m m . . . and yet they are very small and efficient

  4. raffy Says:

    Sorry if off-topic: could goosee.com/puppy be redirected to puppylinux.com and puppylinux.com link directly to puppylinux.com (same way puppylinux.com does now, perhaps as parked domain)? See puppylinux.org for a redirect script (FTP needed).

  5. nic2109 Says:

    Another off-topic post: where have the links gone from this Blog? (Known as the “Blogroll” in WP Admin). They were there at the top RH side yeterday, but have gone today.

  6. cb88 Says:

    depending on your browser they may also be at the bottom….

    can’t the color of SVG icons also be easily changed? perhaps offer to also select the colors that make up the icons….

  7. raffy Says:

    SOS re puppylinux.org. I cannot access admin functions because of this:

    “Error 403

    We’re sorry, but we could not fulfill your request for *URI_hidden*.php on this server.

    Your Internet Protocol address is listed on a blacklist of addresses involved in malicious or illegal activity. See the listing below for more details on specific blacklists and removal procedures.

    Your technical support key is: 7c69-1310-1366-73cd

    You can use this key to fix this problem yourself.

    If you are unable to fix the problem yourself, please contact badbots at ioerror.us and be sure to provide the technical support key shown above.”

    This has happened to me using two of my rather reliable domains, so it’s impossible to go past this screen.

    This is something for servage.net to check, especially for some services where puppylinux.org has been enrolled in the past. Thanks.

  8. nic2109 Says:

    Another off-topic contribution: here’s a very favourable mention of Puppy in PC Advisor. Take a look Barry at http://www.pcadvisor.co.uk/blogs/index.cfm?blogid=4&entryid=1460 and allow yourself a little glow of pleasure!

  9. raffy Says:

    I’ve looked into the problem and this update seems to be the solution:

    However, I still have to find out where Puppian has put those files (the older version). Servage staff (or someone with ssh access) can do it.

  10. mechmike Says:

    Here’s a slew of icons: http://www.downloadpedia.org/Free_Icons_and_Buttons

  11. jeffrey Says:

    Thanks for the alternative icon set Barry.
    The original Dingo’s icons looked rather…
    …well, the new set will look quite professional.
    Much appreciated.

ChmSee CHM help viewer

December 4th, 2007

A few months ago I needed to view a CHM file, which is a compressed HTML-based help file format from Microsoft. The file was the FreeBasic User Manual, and CHM was (and still is) the only format available for download.

There were CHM viewers for Linux, but many dependencies. Until now. I have discovered ‘ChmSee’, which is a GTK2 application and the package uncompressed is 220K and it uses a library ‘chmlib’ that is only 24K uncompressed. It achieves this small size by using the Gecko engine in SeaMonkey. The authors are Yang Hong and Ji YongGang.

It compiled without a glitch, and I downloaded the FreeBasic CHM file and it works perfectly.

ChmSee project page: http://gro.clinux.org/projects/chmsee/
chmlib page: http://www.jedrea.com/chmlib/
Info about CHM: http://en.wikipedia.org/wiki/Microsoft_Compressed_HTML_Help

I think someone recently asked on the forum about Microsoft LIT format — that is a modified CHM format. Here are various related forum threads:


I’ve put this into Dingo and it will be in the next release, alpha3. Note, I have set the ‘application/x-chm’ mime-type in Rox and /etc/mime.types and /etc/mailcap.

One Response to “ChmSee CHM help viewer”

  1. Dougal Says:

    Something unrelated that’s worth looking into:

When the pup_save gets full…

December 4th, 2007

ecomoney provides support for people who use Puppy. These tend to be less computer-technically-savvy customers, and when the pup_save file gets full they do nothing about it, which results in the pup_save file becoming unusable. ecomoney is doing a frugal install for them.

Basically, the customers need to be alerted to the situation — the little ‘freememapplet’ in the taskbar is not enough — and also told what to do about it.

In Dingo I do not have a ‘freememaplet’ in the taskbar, instead there is a message at the top of the screen. The advantage of this is it can contain any text, any font and color. What I have now done is enhance this with a prominent warning, bold black text on a red background.

The warning message varies depending on the PUPMODE. In the case of a frugal install with a pup_save file (normally PUPMODE=12), the warning will advise to delete files or run the ‘Resize personal storage file’ in the ‘Utility’ menu. This message is very prominent so cannot be missed, and is sufficiently intrusive that the user will want to follow the advice.

Note, the user can click on the message to make it go away, but if the free memory drops any lower it will come back. I have set the trigger point at 2MB free space in the pup_save — though, there could be more than one trigger point, with varying severity of warning.

6 Responses to “When the pup_save gets full…”

  1. crafty Says:

    Excuse the fr**ch - ‘about bloody time’ ..(he..he..he..)..

    This has been one of the biggest “newbies” bugbear for quite a long time - so finally getting it sorted out will hopefully ( or definitely - as my better half keeps telling me I should be more positive..hah..) make using Puppy for the “noobs” much easier..

    Great work Barry..


  2. PaulBx1 Says:

    Is the message always visible on top of all windows?

    I sure would prefer to know the trend of things long before I’m down to 2MB. I also use freememapplet to decide when it’s time to clean up emails and such. I like to see 200MB there.

    Maybe keep both the new message for newbies, and freememapplet as well for us long-time users.

  3. BarryK Says:

    The free memory is always displayed at the top of the screen (or it could be elsewhere if that turns out to be better), normally green text with transparent background, so unobtrusive.
    It can be made progressively obtrusive as desired.
    It is not displayed at the top of windows.

  4. cb88 Says:

    perhaps, trigger at 8% of the pupsave is left or 100mb which ever comes first… that would probably be a good comprmise… In any case a percentage based warning would be more helpful/less intrusive since small pupsaves and very large pupsaves are used….

  5. nic2109 Says:

    Sounds like a great benefit.

    I reckon that gentle warnings when below 10% free, strident ones under 5%; with difficult to ignore ones below 2% would be a good scheme.

    Also; don’t forget to use the correct size calculation as per the recent discussions over the size problems in petget…..

  6. ecomoney Says:

    Barry, many (many) thanks for sorting this out, its been a constant thorn in my side for a long time now. Happy this has been drawn to your attention.

Puppy 2.14 Revisited

December 3rd, 2007

Recently in the forum I contemplated updating one of the older versions of Puppy, most likely v2.16.1. The reason for that is that some people experience problems with later kernel and drivers, and some like the combination of applications and small size of a particular older version.

An “upgrade” would mean upgrading to later versions of applications (where possible), fixing bugs, and applying any improved bootup/shutdown scripts, refinements, hardware detection, etc, of later versions. Basically, the objective would be to “smarten it up” to give a closer experience to match recent versions (or even to enhance it in an entirely new way).

Anyway, it turns out that Dougal and pakt have been working on such an upgrade, but to version 2.14. This is not “official” in the sense that I am not involved, so all questions about it have to be directed to those guys. So, if you have a fondness for the older puppies and like the idea of a smaller size (69MB), give it a spin.

This upgraded Puppy is named “2.14R”, where the “R” means “Revisited”. Go to this forum thread for further information:


3 Responses to “Puppy 2.14 Revisited”

  1. Leon Says:

    1.) I manually installed 214R v1.00 using frugal install. As another forum user named ‘fudgy’ mentioned the file searching code at startup works surprisingly fast.

    The step of booting process between this two boot messages:

    ‘Looking for Puppy in hda1… hda2… hda5… hda6… hda7… hda8…’


    ‘Using personal data file /p214r1/pup_save.2fs which is on partition hda5′

    takes less then 1 second. When booting to Puppy 3.01 or 3.92 from the same kind of frugal install it takes about 15 seconds or more.

    2.) I really like the option to mount and to unmount partitions from Rox. This is very nice improvement.

    3.) The Xorg Wizard from Puppy 214R v1.00 solved the problem with missing resolutions in display resolution list.

    We discussed the problem at:


  2. downsouth Says:

    I’ve done a few 2.16.1 remasters - latest being to run an Asus
    eee (typing on it now). Got 800×480 working but I’m using a
    usb flash & usb wireless, as Asus DVD & manual inaccurate.
    Anyway, 2.16.1 has been the most stable for me. My latest iso
    is 132 mb - with all the extras I want except OpenOffice.

  3. downsouth Says:

    … now have a specific Asus eee version of 2.16.1 - 121mb, iso to be installed onto a USB stick and run from there (press Esc at first prompt). Wireless now going. Xvesa only. Available for seriously interested.

ssh-gui ported to GTK2

December 3rd, 2007

ssh-gui is a GTK1 application for making a SSH secure remote login. This application has been in Puppy for a long time, as well as another simpler SSH GUI that I wrote, named “Secure SSH telnet” in the menu.

ssh-gui has a page at sourceforge, however it hasn’t been updated for several years so I guess can be considered a dead project:


Once again, I modified the ‘Makefile’ and this app compiled against GTK2. I have created ’ssh_gui-0.7.1-pup1.tar.gz’ that will be uploaded to puptrix.org soon. This will also be in Dingo alpha3 (as well as Superscan — note, both of these apps are really tiny, only 20KB and 90KB).

Superscan ported to GTK2

December 3rd, 2007

Superscan is a GTK1 port scanner. It has been in Puppy for ages and not compiled for a long time …I couldn’t even find the source. Fortunately, I found it here:


Also fortunately, I only had to modify the ‘configure’ file and it compiled against GTK2. So, this is now in Dingo. I will upload the modified source, which will be named ’superscan-0.8-pup1.tar.gz’, to our source repository at puptrix.org soon. I’m accumulating a lot of new and updated packages for puptrix.


December 2nd, 2007

There was a question asked recently on the forum about how to monitor output to /dev/console while X is running. Many applications will log to /dev/console, which is the console that you see when you exit from X.

The Xorg package has a utility application for that, called ‘xconsole’, and I have added it to Dingo. It is in the ‘Utility’ menu.

One Response to “Xconsole”

  1. Dougal Says:

    New Rox just released.

Xorg vesa working

December 1st, 2007

Dingo alpha1 and alpha2 use the Xvesa Kdrive X server. This is part of the Xorg 7.3 package, but is separate from the “full” X servers. There are two PET packages in Unleashed that I created for this, ‘xorg_base-7.3′ and ‘xorg_xvesa-7.3′.

Now I am incrementally bringing in the “full” X severs. First up is the ‘vesa’ server. Despite the name, this has nothing to do with the Xvesa server. It is similar in that it uses the VESA video modes and is not hardware-accelerated. It differs from Xvesa in that it can use the various modules and extensions that are in the Xorg package (whereas Xvesa is a standalone product) — for example, the Synaptics touchpad driver.

I created another package, ‘xorg_xorg_base-7.3′, that has extra stuff needed to support the full X servers, and I put the ‘vesa’ server in it. I added this package, and now Xorg runs. It’s quite small, only adds about 1.3MB.

…like kirk, I’m very happy to have the Synaptics driver! (thanks for compiling that!)

While working on this, I discovered that the Xorg Wizard has a bug. A rounding-off error prevented my laptop running at 1280×800 and it ran at 1024×768. Fixed.

Next step, I’ll create packages for specific X servers, or maybe one package for a bunch of them.

6 Responses to “Xorg vesa working”

  1. lobster Says:

    Good news :)

    The one bug developers always miss and dismiss (be honest they do not even regard it as a bug) is ‘look and feel’.

    It is the first one users see.

  2. raffy Says:

    Happy to hear about this basic package. For embedded systems, the usual chips are the AMD Geode, VIA Unichrome, and an assortment of SiS chips (see the Intel Classmate, eee and D201GLY2 board as examples).

  3. kirk Says:

    Maybe just one big add-on package with everything. Would need the DRM modules as well if they’re not going to be in the zdrv. May also want to include a new /etc/X11/xorg.conf0 that has :

    Load “dri”

    in the Section “Module”.

    The extra Megabytes of package size would probably be worth it for the simplicity. Probably be a pet package ~25MB.

  4. tempestuous Says:

    I installed the Xorg 7.3 package, and then proceeded to compile the latest MPlayer 1.0rc2.
    I had a lot of trouble with this compilation until I realised that the configuration script was failing to find libraries in /usr/X11R7/lib
    This command fixed the problem

    export LIBRARY_PATH=/usr/lib:/usr/local/lib:/usr/X11R7/lib

    Can anyone tell me how I permanently set this, so I don’t have to run that command each time I power up?

    Barry, I made some suggestions about Puppy’s version of MPlayer on the main forum -

  5. kirk Says:

    I think that’s in /etc/lld.so.conf. Think you need to execute ldconfig after editing that file.

  6. BarryK Says:

    kirk: ok, have put in ‘load “dri”‘
    tempestuous: I think in /etc/profile.

Busybox updated

November 30th, 2007

There is a recent forum thread about the ‘adduser’ applet in Puppy not working. This is for Puppy 3.01 I think, which uses Busybox 1.6.1. Earlier puppies used a separate package named ‘tinylogin’, and in the forum discussion it was found that the ‘adduser’ from that old package does work.

I found mention by the developers that adduser has been fixed in Busybox 1.7, so I have upgraded Dingo to the latest Busybox, which is version 1.8.2.

Note that the initial-ramdisk (initramfs) still has an older Busybox, v1.4.1 I think, compiled statically with uClibc. That seems to be okay so won’t change it.

2 Responses to “Busybox updated”

  1. BarryK Says:

    Does anyone have a email address for Hacao?
    I sent him a p.m. on the forum a few days ago, but he hasn’t logged in. I’ve got some questions about the Classmate!

  2. caneri Says:

    hacaolinux-AT-gmail-DOT-com should work.

    That’s how I stay in touch.


Pmount rewritten

November 29th, 2007

I have completely rewritten pmount, the Puppy drive mounter.

You can find it here:


It is file ‘pmount.gz’. Gunzip it to /usr/sbin — it must be placed at that directory and must be named ‘pmount’ — rename the old one to hide it.

I have only tested this in Dingo alpha2.

To test mounting an NTFS partition, you have to get rid of /bin/mount — rename it to something else. You will see /bin/mount-FULL — rename that to ‘mount’.

I don’t have MUT in Dingo, but it was mentioned in a thread today where we have been discussing bugs in Pmount, that MUT mounted a NTFS partition read-write — one thing to be careful about though: if an NTFS partition is mounted writable by using the kernel NTFS driver, writing is not safe. The ‘ntfs-3g’ FUSE user-mode driver is required, and you can check whether this has been used by executing:

# mount

after the NTFS partition is mounted, and you should see the string “fuse”. If not, it is unsafe to write.

Testing of the new pmount invited.

19 Responses to “Pmount rewritten”

  1. plinej Says:

    Works fine for me, although I didn’t have any problems with the old one.

  2. boscobearbank Says:

    Smaller is better. However, I get an ntfs-3g error when using the new pmount (and the old one too):
    The ntfs-3g driver was unable to mount the NTFS
    partition and returned this error message:
    shell-init: error retrieving current directory: getcwd: cannot access parent directories: No such file or directory
    mount: illegal option — i

  3. lluamco Says:

    Using the new pmount, making it executable in /usr/sbin/pmount and renaming /bin/mount-FULL to /bin/mount, I get the following in puppy 3.01:

    When mounting an NTFS partition, rox opens its contents, but pmount does not register it as mounted. Also, when executing mount, “fuse” is not displayed. However, if an ext3 partition is mounted (pmount registers it OK) and unmounted, then the previous NTFS partition is registered as mounted and, execution of mount displays “fuse”.

  4. hairywill Says:

    Works for me, but then it did before. Much more compact, if there are any bugs I’m sure they will be easier to find. I notice you’ve gone back to opening rox on a successful mount, I prefer this. You might consider these two edits to get the rox window to pop up in front of pmount. The first line I’ve included in each case is your code for context.

    echo “$PMOUNTGUI” > /tmp/pmountdlg.txt
    if [ -f /tmp/pmount-launch-rox.sh ]; then
    source /tmp/pmount-launch-rox.sh &
    rm /tmp/pmount-launch-rox.sh

    [ $RETVAL1 -ne 0 ] && xmessage -bg red -center -title “ERROR: unable to mount $DEVNAME”
    [ “$RETVAL1″ -eq 0 ] && echo ‘#!/bin/sh
    while [ ! “$PMOUNTUP” ]
    PMOUNTUP=`ps | grep “gtkdialog3 –file=/tmp/pmountdlg.txt –center”`
    usleep 100000
    rox -d “/mnt/’$DEVNAME’”‘ > /tmp/pmount-launch-rox.sh

    I don’t expect you to lave my name in :-)

  5. BarryK Says:

    “Smaller is better. However, I get an ntfs-3g error when using the new pmount (and the old one too):
    The ntfs-3g driver was unable to mount the NTFS
    partition and returned this error message:
    shell-init: error retrieving current directory: getcwd: cannot access parent directories: No such file or directory
    mount: illegal option — i”

    That would happen if you didn’t follow all instructions regarding renaming on ‘mount’ applications — did you do that also?

  6. BarryK Says:

    Luamco, I cannot reproduce your problem. One thing, in 3.01, do you have /sbin/mount.fuse.ntfs, mount.ntfs-3g and mount.ntfs-fuse symlinks? …I don’t think that will make any difference though.

  7. boscobearbank Says:

    Barry, it’s not a renaming thing. I followed the instructions as best as someone who’s philosophy is “if all else fails, RTFM” can. I get the same error when I use my pristine Dingo alpha 2 CD and pmount or ntfs-3g itself.

  8. BarryK Says:

    I have just updated the pmount script on ibiblio.

    A couple of bug fixes where xmessage might have caused a crash, now can run multiple instances of pmount, now the Rox window displays on top when a partition mounted (thanks hairywill).

  9. jb4x4 Says:

    THIS ONE WORKS!!! - Thank you !! (running v3.01) To better explain myself — I have a 15 partition drive with hda1 as a 100M boot partition. It always showed up as “.G” and would not mount - only showed an error of being a removable drive. This one works fine, shows correct size and mounts. BTW, nice touch with auto opening rox and the speed improvement.

    I also tried NTFS mounting, WITHOUT changing /bin/mount. The drive mounted and opened rox. I did not try any reads or writes though.

  10. BarryK Says:

    boscobear, I found that the newer ‘ntfs-3g’ calls ‘mount’ internally, and as I had a script named ‘mount’ that also calls ‘ntfs-3g’, there was a weird situation. I have rolled back to the full mount program and that fixed the problem for me.

    If you type ‘which mount’ in a terminal, does it find the full mount program?

  11. kirk Says:

    Works for me.

    I tried vfat, ext3, and ntfs. My ntfs was hibernated and so it was mounted read only, just as it should. I did get two red waring windows with different messages, though they both said about the same thing about the partiton be mount read only.


    The init script will fsck my ext3 partition which adds 3-4 minutes to the boot time. I don’t think there’s any reason to fsck a ext3 partition.

    My ext3 partiton has been crashed countless times, sometimes because my battery dies, or a new game I’m trying locks-up. or I just decide to delete the pup_save file I’m using and then just hit the power switch. After about 6-8 months of doing this I ran a fsck on it:

    # e2fsck /dev/sda4
    e2fsck 1.38 (30-Jun-2005)
    /dev/sda4 has been mounted 321 times without being checked, check forced.
    Pass 1: Checking inodes, blocks, and sizes
    Pass 2: Checking directory structure
    Pass 3: Checking directory connectivity
    Pass 4: Checking reference counts
    Pass 5: Checking group summary information
    /dev/sda4: 89516/1938272 files (3.9% non-contiguous), 2322368/3899392 blocks

    As you can see fsck had nothing to do.

    This is the first time I’ve been able to use Dingo much (thanks for the Xorg tarball). Seems very snappy! I really think your on the right path with T2. Looking forward to the Dingo - T2 build package.
    Keep up the good work!

  12. BarryK Says:

    jb4×4 wrote:
    “I also tried NTFS mounting, WITHOUT changing /bin/mount. The drive mounted and opened rox. I did not try any reads or writes though.”

    After it has been mounted, would you mind running ‘mount’ in a terminal and see if the string “fuse” is there in the ntfs entry.

    Then, test with the full /bin/mount program (replacing my mount script) and see if ‘mount’ returns the same info.

    …cause, I’m puzzled. I had to get rid of my ‘mount’ script to get rid of the weird error message and to get it to mount read-write.

    I’ve just had a thought what the problem could be. I think that my NTFS partition is “dirty”, whereas yours is probably “clean”, which affects how ntfs-3g handles it — which from previous experience does change with different versions of ntfs-3g.

  13. jb4x4 Says:

    Result WITHOUT changing mount–

    /dev/hdb1 on /mnt/hdb1 type fuseblk (rw,nosuid,nodev,noatime,user_id=0,group_id=0,default_permissions,allow_other)

    After changing mount-full to mount– (original mount changed to mount1)

    “nothing” - (all other partitions still listed)
    Rox still auto opened with the filesystem, but pmount showed hdb1 UNMOUNTED.

    Note: I don’t believe I have ever tried writing to my NTFS partition with Linux (call it paranoia, I don’t want to bork it :) )

  14. jb4x4 Says:

    This is strange…. Still working with mount-full. I reported that pmount reported the NTFS partition (hdb1) as unmounted even after rox opened. If I mounted the save partition on said drive (hdb2, vfat) and unmounted it. When pmount refreshed, the NTFS partition (hdb1) showed as mounted…..

  15. BarryK Says:

    Hmmm, I think I know what is going wrong. When ‘mount’ is executed without any parameters, which my pmount script does also, it returns the mounted partitions. However, the full ‘mount’ reads /etc/mtab, which perhaps is not fully up-to-date. The Busybox ‘mount’ ignores /etc/mtab and instead reads /proc/mounts, which is accurate.

    So, we can both try this: leave the full version as ‘mount-full’, rename my script as ‘mount-HIDE’ and rename ‘mount-BB-NOTUSED’ to ‘mount’.
    Do the same to the ‘umount’ application, that is, use the Busybox ‘umount’.
    Then, replace /etc/mtab with:
    # ln -s /proc/mounts /etc/mtab

  16. BarryK Says:

    Ignore my previous post!

    I have just understood things a bit better. The fundamental problem is that the ‘ntfs-3g’ program calls ‘mount’ — and if it the Busybox mount, passes a parameter that mount does not understand. That is the crux of the problem.

    One solution is to use the full mount, full umount, and leave /etc/mtab as a file. The full mount and umount will keep /etc/mtab up-to-date.
    This solution has a problem though, as if a script uses the Busybox mount/umount (by executing ‘busybox mount ….’) then /etc/mtab will not get updated.

    Another solution is that I fix my ‘mount’ script to recognise the passed param that is causing the problem. I’ll check that out now.

  17. BarryK Says:

    Ok, I think it is fixed. I improved my ‘mount’ script. I’ll leave it at that, the fixed ‘mount’ and ‘pmount’ will be in alpha3.

  18. BarryK Says:

    I’ve uploaded the latest ‘mount’ script to ibiblio. Download that to /bin/mount.

    No other changes to alpha2 should be needed. Just the pristine alpha2, with the updated /bin/mount and /usr/sbin/pmount scripts.

  19. lluamco Says:

    Barry, the last versions of mount and pmount taken from ibiblio work OK. Thanks.

Complete Xorg binary tarball for Dingo

November 29th, 2007

For whomsoever might be interested, here it is:


I compiled this in T2. It’s the complete Xorg, all the drivers, headers, etc. I’m wondering what is the best way to break this up into PET packages …maybe a base package, then packages for each type of video hardware? Or, create packages like we have in earlier puppies? If anyone, such as Dougal, feels motivated to play with this, go for it. This tarball expands into its own directory ‘xorg_complete-7.3′.

10 Responses to “Complete Xorg binary tarball for Dingo”

  1. raffy Says:

    Thanks, Barry, Yes, Dougal does wonders to the video wizard.

    It must be good to have a package for the Geode (GX466, LX800). For companion chip, it’s the CS5536 that is now being used. (Geode is used in the OLPC, specifically the LX700.)

  2. kirk Says:


    I really haven’t been able to use Dingo much because of my touchy Synaptics/Alps touch pad. Drives me crazy. I found a patched version of the Synaptics driver and compiled it in Dingo. My sanity has returned! I posted the source and a binary here:


  3. kirk Says:

    Tied to get DRI going in Xorg. All seemed OK until I tried glxgears:

    # glxgears
    glxgears: /usr/lib/libstdc++.so.6: no version information available (required by /usr/X11R7/lib/libGLU.so.1)
    glxgears: /usr/lib/libstdc++.so.6: no version information available (required by /usr/X11R7/lib/libGLU.so.1)
    Broken pipe

    Did some googling, similar problems seem to come from using a different version of gcc than what the library was compiled with. Also I’m using the DRM modules from 3.01, assuming it’s the same kernel.

  4. BarryK Says:

    “/usr/lib/libstdc++.so.6: no version information available”
    kirk, yes, you’ll get that message for any C++ app not compiled in Dingo. For me, it hasn’t stopped an app from running though.

  5. kirk Says:

    Maybe I’ve got something messed up. Xorg is working. The log looks good, but the glxgears screen is blank. I’ll keep working on it. If someone gets DRI working please post.

  6. kirk Says:

    I compiled ATI’s latest fglrx driver in dingo, now all is well with DRI. The open source driver worked with my ATI X300 Mobility card in Xorg 7.2. Maybe it’s not supported in 7.3? The ATI driver does replace libGL with it’s own version and build it’s own DRM and DRI modules. Surprised the fglrx driver built with the kernel/gcc mismatch. But I’ll take it! Sorry to spam your blog.

  7. Dougal Says:

    “I compiled ATI’s latest fglrx driver in dingo”

    How could you do that? ATI didn’t say they’ll be opening the code for it, just supplying specs for new free drivers to be written.

  8. tempestuous Says:

    With the proprietary ATi and nVIDIA drivers, just the DRM kernel module is provided in source code form. All other Xorg driver components and libraries are supplied in binary form. The installer script just checks what version of Xorg you have, then installs all binary files into the correct locations.

  9. kirk Says:

    tempestuous is correct, just fglrx.ko is compiled everything thing else is a binary.

  10. nic2109 Says:

    Any chance of placing the compiled ATI driver plus the other binaries somewhere as a package, Kirk? Those of us with one of their cards need it, and we haven’t your skills!

JWM is cause of MPlayer problem

November 28th, 2007

All versions of JWM have a bug that causes a window to draw with the titlebar off the top of the screen. This only happens sometimes, rare enough not to be a nuisance, however it is happening much more when I am running in the 800×480 screen of the Classmate.

Starting MPlayer in Dingo, sometimes only one window displays. There are supposed to be two, the control panel and the video output. It’s strange, like a lottery which one displays. But, the missing one is still there, just off the screen, and right-click on taskbar can be used to move it onto the screen.

I tested with Openbox, and MPlayer windows display perfectly.

So, I have notified Joe, the JWM developer.

The reason I decided to test Openbox is because Nathan chose it for his Grafpup v2, so I reckoned he must have had good reasons. Does anyone know what panel application he has used? Is there any GTK2 panel/taskbar application?

I’m thinking that if Joe can’t fix the problem, I’ll follow Nathan and move to Openbox and a suitable tray/taskbar panel.
…I hope Joe can fix it though, as we have invested much effort in JWM, themes, config tools, plus familiarity with it.

21 Responses to “JWM is cause of MPlayer problem”

  1. Leon Says:

    The first thing that I’ve done after installing every new version of Grafpup was that I installed JWM. I also tried IceWM and some others window managers but didn’t found anything so simple and effective as JWM. Is it worth to change window manager because of the problem with starting MPlayer?

  2. disciple Says:

    Nathan used an old version of Lxpanel (of LXDE) that could use XDG menus (which current versions cannot). It is GTK2.
    Personally, I prefer JWM, but that is a pretty nasty bug. I’m lucky - I think I’ve only had it happen twice.
    Openbox and Lxpanel do seem a good fit with the GTK2 philosophy of Dingo :) But I think they are much slower on older machines :(

  3. disciple Says:

    BTW you might want to check Nathan’s blog. I seem to remember he was talking about doing something different for menus or taskbar for some reason.

  4. fwiffo Says:

    > Is there any GTK2 panel/taskbar application?

    I do know about

    - fbpanel: http://fbpanel.sourceforge.net/index.html

    - lxpanel: http://lxde.sourceforge.net/

    and just discovered

    - foopanel: http://foopanel.berlios.de/foopanel/

    which not only depends on python libraries (so isn’t as lightweight as the others) but also seems to be in its infancy.

    Finally I just discovered a really tiny panel, though without menus, launchers nor applets:

    - fspanel: http://www.chatjunkies.org/fspanel/

    Hope this helps.


  5. Henry Says:

    Here’s a problem I’ve noticed only through the 3 and 4 versions. Maybe it’s just me - I haven’t seen it in the forum - or maybe it’s a JWM problem ;-)

    I always keep a number of different personal configuration files for the usual reasons. It seems I used to be able to rename, move around, these files without any problem. Now it seems I have to walk on eggshells to avoid losing my pinboard customizations. If I rename a config or if I add or remove an sfs file, e.g openoffice, the pinboard has to be rebuilt. Surely this is not how it’s supposed to work? I have worked around it by keeping all pup_save.2fs files named just that but keeping them each in it’s own folder. True, you have provided a means to individually name a config on first save, but I still have the problem on subsequent incremental configurations. I suppose I could have said this in fewer words :-)

  6. BarryK Says:

    Well, probably Joe will fix JWM, …eventually.

    Openbox is nice, it has a GUI configurator which works in Puppy. I’ll make these into PET packages.

    A strange thought — maybe JWM could be hacked to only display the panel, and Openbox used as the window manager.

  7. kirk Says:


    Nothing to do with JWM. Just wanted to mention that the link to this page from here:


    points to the old blog.

  8. cb88 Says:

    jwm panel only…. with swm for the window manager…

    better window manager switching….really needed DSL has it already

    couldn’t it be similar to the screen lock configuration?

    the config would have to restart X and pass the correct WM to xwin…..

  9. Dougal Says:

    It might be worth to look at the latest fbpanel — LxPanel is based on fbpanel bus hasn’t been updated in over a year, while fbpanel had a couple of releases a few months back.

    I’ve run Puppy with Openbox+LxPanel. The only problem is that you need to “standardize” the Puppy XDG system to use LxPanel — it has an internal XDG engine and won’t show most of the Puppy apps…
    Also, Openbox doesn’t support icons in the menu.

  10. BarryK Says:

    kirk, thanks, I fixed the page.

  11. MarkUlrich Says:

    I will try to add the functions required to display menus in GtkBasic in the next 2 weeks.
    Then the creation of a Gtk2 startmenu will be a breeze.

    More difficult would be a taskbar.
    One problem you have if you use several applications is, that Ram-usage increases dramatically.
    Also the integration of WM/Taskbar is very efficient, as windowhandling can happen directly, while for seperated apps you must define a protocoll.
    Xlib has mechanisms for that, but doing it in a single app certainly is more efficient.

    Btw., what about icewm? Most puplets use it by default.


  12. BarryK Says:

    MU, yes, MPlayer works fine in IceWM. How do puppy-ites configure IceWM? — as the gui tools I found require Python — do they edit the config files manually? Do you have an opinion who has done the best puppy-isation of IceWM? — pizzagood with pizzapup?

    I’m sure that Joe will fix JWM, he always has fixed problems before. And, I think most of us have gotten to prefer JWM over anything else.

  13. BarryK Says:

    The recent versions of lxpanel and fbpanel both use XDG. Perhaps Nathan went back to an older version, (v4.5 of fbpanel is the last that did not use XDG, latest is 4.12), that does not use XDG as he wanted to manually set the menu.

  14. wolf pup Says:

    in mplayer, under preferences, misc tab, theres a setting that may be unchecked about save window position and show video window when inactive could be unchecked to only show when video is played.

  15. MarkUlrich Says:

    The best implementation might be ezpup by Whodo.
    But it is rather huge.
    There was written a configtool once by a forum-member, Icewinconfig.
    It is included in ezpup.
    Looks like it was done in MurgaLua.
    I plan to write a configtool in GtkBasic for Muppy, if you plan to switch to icewm, I will give this high priority.
    Else it would be done for the next release muppy009 in some months.

    I’m away for some days from tomorrow on.
    Hopefully I will get my own internet on tuesday.
    We could start a thread in the forum for such a Gtk2 based icewm-tool.
    Where we could write down, which options it should include.
    I think it makes no sense to add all options from the preferences file, as those are overwhelming.


  16. raffy Says:

    May I add that a way for manually modifying the menu be added to Puppy (whichever WM is chosen), much like the way the backgroundsetter or “Set Desktop Icon” work: drop a file into a category and the menu item is created; dragging an item up or down can change the order of the items. (XDG is convenient for adding programs, but the user may want direct control over his/her menu.)

    Thanks in advance to the coders who’ll do it. :)

  17. disciple Says:

    “The recent versions of lxpanel and fbpanel both use XDG. Perhaps Nathan went back to an older version, (v4.5 of fbpanel is the last that did not use XDG, latest is 4.12), that does not use XDG as he wanted to manually set the menu.”

    That’s interesting, - when Grafpup 2 came out XDG was definitely disabled in the current version, so Nathan used an old version:

    He seemed to be having all sorts of problems with both fbpanel and lxpanel, so I think we’d want to test them quite thoroughly.

  18. tempestuous Says:

    What about the Oroborus window manager?

  19. MarkUlrich Says:

    yes, Oroborus is nice.
    2 arguments against it:
    someone wrote, it is slow. But bombayrockers wrote, it is fast.
    So it has to be tested on old hardware.
    The other: not much development.
    The last release is from 2005.
    As especially the NetWM specifications by freedesktop.org might change, it is important to have a WM that is updated often.

    I started writing a tool to configure icewm.

    Basically works with checkboxes, thats to say 0/1 values.

    I also need strings/numbers like xxx seconds.
    I don’t want to put too much work in a tool just for icewm, so I started on a config-center, that shall read an own type of configuration (you easily could convert the one from icewm then).
    This would allow to use it also for other settings.
    I then encountered some bugs in the script, that generates GtkBasic.
    This prevents some functions I need from being built.
    So I will fix this first, I already did a new rewrite, maybe 60% are done.

    Then I can continue on the configcenter.

    In GtkBasic, I also can drag around with the mouse now borderless windows.
    So I could build an own pinboard :)
    Unfortunately this takes longer, as it is difficult to render text under each icon. This also is a reason for the rewrite of the GtkBasic buildscript. I hope, then the required pango functions can be generated automatically.
    Expect the configcenter maybe this weekend, a pinboard-prototype in a month or later.
    I will have 2 weeks full of work before christmas, so no time for a pinboard itself at moment.


  20. tronkel Says:

    I wrote the icewinconfig manager as used in EZpup.
    It was done using a standard FLTK interface and implemented in C++. There exists both a GUI and command line version. They are both fairly small. I wonder how the size and speed of execution of the program would compare with something written in GTK Basic

  21. MarkUlrich Says:

    there is a prototype available.
    It currently just uses an own configfile, so a converter icewm -> config.rc -> icewm must be added.
    I think I can release a version shortly before new year.


    The speed seems to be very fast, though I just can test it on a newer computer at moment.
    But in general, GtkBasic does not need many resources, maybe twice as much as leafpad.


Puppy Dingo 392 alpha2 available

November 26th, 2007

I’ve been busy, you need to read the flurry of recent posts to this blog!

Get this new Puppy from here (iso is 77.8MB):


Everything is still in a state of flux.

We have the new PCMCIA system for 2.6 kernels, but it is broken.

I have included ‘flac’ but some apps that could make use of flac have been compiled without it, which will get fixed sometime (RipOff for example).

The Universal Installer still won’t install to USB. I haven’t gotten around to that yet.

Things of special interest:
The full Inkscape!
CUPS rollback, needs testing, maybe fixing, involvement invited.

There are still lots of things to improve… well this is an alpha. The usual warnings, do not upgrade an existing pup_save, this alpha is not yet ready for general release.

32 Responses to “Puppy Dingo 392 alpha2 available”

  1. Sage Says:

    Strange new version numbering sequence? Appears to be 4.0alpha2.
    MUT still missing. Still need an aircraft sick bag at hand or pair of sunglasses. At least the wallpaper swapping seems to work now!
    Lacks the refinements that make a user’s life easier. For example, in DSL, when the NIC IP address is entered, guesstimates of all the other static IP parameters are automatically inserted. The GW, broadcast and netmask will almost always be guessed correctly, leaving just the DNS of the nameserver to be edited. I’m sure Barry feels this is trivial - not so if you use Puppy in it’s originally intended liveCD mode. The main preoccupation will be to get online ASAP. Save the settings on exit? How to do that if no HD present? For distro junkies, would need to add a Puppy .sfs along with the other 200 distro files to an HD.
    Furthermore, perhaps it’s a nice courtesy to open the CD tray on exit. Klaus K offered this one way back in the cycle of liveCD development. Such a nuisance having to restart and re-stop machine or search for a paper clip to liberate the CD.
    There will always be a dichotomy of priorities between users and developers. Attention to detail may be less intellectually stimulating but, in the longer term will attract a bigger circle of users, providing more input, allowing developers better feedback and a wider range of ideas and suggestions to engage their evident talents.

  2. veronicathecow Says:

    Hi Barry, I am trying to download this but for some reason it keeps stopping partway. I installed “Down them all” download manager for Firefox. When I needed to stop it is said that this download was not resumable. Anyway can it be made resumable from your end?
    I would imagine this would reduce bandwidth as there would be less failed downloads and more happy downloaders. Cheers

    P.S. Re compiling in RAM, I just brought some sticks, do you need any and if so what type and if I have them I will send you some free of charge as a thank you.

  3. Henry Says:

    Thanks, Barry!

    Lots to explore here. Of course I’m writing from alpha 2.


  4. cb88 Says:

    One way to compile faster is to use the -pipe flag which uses pipes instead of file i/o when compiling may even speedup compiling in ram….

    I am not at all happy with puppy right now…. what do you mean i can compile modules in 4.00 what is the point!?

    gktmm …. thanks for giving the OK to more gtk bloat….since it is built ontop of GTK I expect an exponential decrease in proformance…..

    where are the bugfixes?

    where are the imrovements?

    As far as using the same kernel on the classmate… I think not. There will probalby come along a puplet to “fix” the support the “real” “classmate puppy” should have to begin with i realize that the kernel won’t be gimped but it won’t be the best either. It has a limited hardware range and a special kernel for it is ideal….

    Puppy used to let you take out the CD what happened?

    Puppy does not meet its goals and that is a shameful fact because it CAN.

    The installer(PUI) does not work half the time and petget is just wierd …. every package install i have done in the past few weeks petget has borked (it will just copy the package to / and act like nothing happened and i have plenty ram and have done the space checking fix)(this is a reoccuring bug and has happened with multiple pupsaves an should be extremely easy to replicate)

    I have been experimenting with lzma-utils in order for it to work the files must be compressed individually then tared up…. not the other way around makes for extremely good compression ratios 0.31 on the abiword binary for instance

    I know this post is a little harsh but i wish you’ed take a look at your mission statement and see (for real) how well you are sticking to it…..

  5. pakt Says:

    Guys, remember this is just an ‘alpha’ of a major Puppy re-write, released by Barry to help him test and get it working. So we shouldn’t expect all the details to work yet ;-)

  6. raffy Says:

    Oh, boy, cb88 is really stressed. ;)

    Maybe he can take on customizing for the Classmate, for a change. :) Or pakt…

  7. pakt Says:

    Just tested CUPS wizard on Dingo-a2 - no problem getting my Brother HL-2030 to work (didn’t even ask for username/password) :-)

  8. Sage Says:

    Better that Barry gets all the nit-picking during alpha than minutes after release of the final.
    Yes, most incisive cb88, ‘mission statement’ was the phrase I was searching. Clarity of purpose may not be the primary consideration of a partly retired hobbyist, but it would sure help his acolytes and camp followers! Remember : only your very best friends will give you unvarnished criticism without rancour or malice. Not all of us are able to come up with helpful suggestions and assistance, though……..

  9. pakt Says:

    Sage, I think you put your foot into it….

    By definition (http://www.thefreedictionary.com/nitpicking):

    nit·pick·ing (ntpkng)
    Minute, trivial, unnecessary, and unjustified criticism or faultfinding.

  10. raffy Says:

    Using Alpha2 now. Immediately enjoyed the new features of Inkscape.

  11. flagman Says:

    I’d like to remind everyone that we’re dealing with a second Alpha here. It’s still basically a proof of concept release. So let’s take it easy. Barry will certainly fix what’s broken. What he needs most now are impressions, suggestions and guidance.

    Thanks Barry, for thinking outside the box and sharing this early work.

  12. cb88 Says:

    @raffy heh, I have just begun to figure out how to compile a kernel…. just yesterday compiled 2.6.22.something with practically all the drivers in kernel… perhaps useful for netbooting like you do raffy it came out at 15.8mb….haven’t gotten it to boot all the way though I’m getting “kernel not syncing : VFS: something about not being able to find root on block (2,0) NO, i am not using a puppy HD install i am attempting to build a minimal linux system around the latest buggybox (I formerly knew thee as busybox) :-)

    It seems that every new feature is built on its own library totally disregarding that it might be better to extend exsisting libraries….

    I’m not really against puppy or barry as far as the bloat that has occured in the applications… (but i am really pissed at the devlopers of said apps)

    however, I am dissapointed that the bugs that ARE in puppy haven’t been fixed… It has come to the point that many puppy users are rewriting the core components of puppy ie( xorgwizard….very good but still needed a little work) and leachim has mentioned rewriting how puppy saves files…several puppy users are rebuilding kernels with long awaited features and the list goes on…….

    personally I am a little embarased to even hand out puppy CDs because it doesn’t JUST WORK …. I do think 2.14-2.15 is a pretty good though

    It is not that i am stressed at all I can work around bugs like petget being broken and PUI (I just don’t bother anymore) but i am a computer engineering major fro crying out loud its what i do! You can’t expect regular people (yes i mean people with masters degrees and doctorates) to be able to fix these things becasue it is not thier field of expertise. I imagine that they could figure it out eventually but the only reason they are is because we are shipping broken features and hardly ever fixing bugs its not about going back and fixing all the old versions it is about getting it right in the first place….

    Perhaps I am a strage sort of perfectionist (not in all areas notice the lack of capitaization….) but that is what programmers and engineers should be. I might not pickup all my mistakes on the first draft of a paper but that doesn’t mean I’m just gonna leave it that way. I’m going to get it right before the final…. I don’t see how anybody let that petget bug get through the alpha stage….seriously

    @pakt: I realize this is an alpha that is why I an writing this long winded post to begin with…. :-)

  13. Sage Says:

    Hey, pakt, hur jor ni?
    Your English grammar & comprehension may be impeccable, but you’d need to live here to appreciate that contemporary usage of nit-picking, whilst being minute or trivial, no longer implies unnecessary or unjustified, nor even critical or faultfinding. C’est la vie.
    And all this whilst our unsuspecting Leader sleeps soundly.

  14. Dougal Says:

    Where did the 392 come from?

    Anyway, for the new PCMCIA system to work you need to enable it in the kernel:

  15. bios007 Says:


    I used Puppy on my old notebook(Gericom k6/2 500 96mb), and i love it!
    But this new version is FANTASTIC!!
    Excellent running,mutimed perfect(with mplayer) ,wlan super.
    A few suggestion for development:
    -usb eth now not working.until this was ok..(with 3com 3c59x driver)
    -bluetooth support neeeeeed!
    -i think openoffice for default office is better, like other…things.
    -full multimedia support for default.(codecs)
    -firefox for default browser
    -thunderbird for def. mailer
    -forget the businesscard cd-size, its over.i have many problem if i want buy a mini-cd! and this is 210mb.
    Now thats all…but others Grrreat!


  16. cthisbear Says:

    veronicathecow Says: “it keeps stopping partway.”



    Boots quickly on old HP 2.0 Gig.

    “Puppy used to let you take out the CD what happened?”

    OK on test HP 772A with 256 ram.
    Still not working on NTFS.

    Sorry Barry but this is crap.
    Time to fix it.

    Otherwise John Howard event bound to happen.
    The red screen of death.
    Now frothing at the mouth….Puppy is so good at rescues…till Puppy 4.0.

    Internet connection great…extremely quick recognition.

    I know whingeing is easy.
    Creativity and its implimentation is frustrating, and rewards are few.
    You have the brilliance…..but a few globes need changing.


  17. BarryK Says:

    A couple of the replies to this thread are most inappropriate. A Puppy ‘alpha’ is not a release candidate, it is usually for me to make available some recent developments for testing. Some of the above posts are treating it as though it is a candidate for release. Perhaps I should use a different Greek letter? What comes before ‘alpha’?

    Some of the criticisms are very vague. “Sorry Barry but this is crap” is so inappropriate. The post doesn’t even explain why. The NTFS problem mentioned, existed before, not introduced by this alpha. The NTFS problem, that some people have, is not even Puppy’s fault.

    Regarding bloat, the Inclusion of Inkscape does not mean it will be in the final release. Quite likely it will be a PET package (as usual). I included it in this alpha for testing.

    The entire spirit of testing these alphas is being subverted by a couple of people, and they are now warned. I need feedback that will help me, such as Raffy’s testing of the printing — thanks for that, it looks like we have a goer there! — anyone test foomatic? For those who just post to whinge and not really contribute, I can do without that, and you will be banned from this blog — you can whinge as much as you like over on the forum.

    Dougal, I used version 3.90 for alpha 1, 3.92 for alpha 2. This is just for the convenience of keeping them apart, so devx files and so on don’t get mixed up. Thanks for the info on the kernel parameter for PCMCIA, it looks like I’ll have to recompile the kernel. I also need to do a recompile in T2 with all the Xorg servers, and I’ll get onto that as well.

  18. BarryK Says:

    Tronkel has started a forum thread for feedback on Dingo alpha2:

  19. cb88 Says:

    I guess it is time I stepped off the high horse a second and just say that I am sorry if I have offended you Barry I really didn’t mean to come across like that.

    I even made a point to say that most of the biggest problems are Linux problems in general when I said that i am not very happy with thier

    developers (mozilla others etc…) not even that I don’t appreciate thier goals just well…..

    I am not against inkscape as long as the price is not too high and you seem to think the same way….

    I realize this is an alpha …. a good time to look at what we have a make changes….as you are doing

    Barry if my criticism was vague I am sorry the biggest problem i have with anything that you have developed is petget i like it when it works but something is wrong with it… if there is any tests or things i could do to help debug tell me…

    As far as the petget bug or whatever it is perhaps it is actually the packages fault? if the package were somehow misspackaged would this happen? I know wow’s compiz (older package not sure about the new one) package used to mess up petget for sure…from my standpoint it is an “old bug” but perhaps you haven’t seen it

    Please don’t think that I don’t understand that you do try to get it all working … amid the fast pace of puppy development in some areas it just sometimes seems that others areas are at a stand still.

    As far as booting up and doing regular stuff puppy does well it is just those extras…that throw a wrench in the works..

    I realize I make a lot of posts about things not working or how something is bloated as a sea cow but if everybody just went along and didn’t bother to say something would it ever get any better?

    I just got a chance to download 4.0a2 and I’ll see if i can figure out how to reproduce the petget bug in it

    while you are speaking of xorg have you heard of any way to fix the neomagic support? I think i read that the screen goes corupted because the chips is not reinitialized or somthing… only happens when x is killed using xvesa

    I really am sorry for the contributing to the subversion wish I could have though of how to say it in a better way

    I think i had better be quiet now (as if i hadn’t wasted enough kb in this thread already)…. have fun getting it all working…

  20. grumpywolfe Says:

    Hi BarryK

    “The Universal Installer still won’t install to USB. I haven’t gotten around to that yet.”

    I created a login to let you know that I am running puppy 4.0 alpha 2 on a usb thumb drive. I have cups running my HP deskjet 940c. The printer is on usb.

    Thanks for all your work I like puppy

  21. disciple Says:

    “where are the imrovements?”

    The whole idea is an improvement :)
    GTK2 only; run in 128MB ram…

  22. BarryK Says:

    With PETget, you have to be specific about what is the fault. I did a bugfix between alpha1 and alpha2 that may have fixed the “disappearing packages” in the 2-pane window.
    However, PETget will not actually download anything, as I have setup a new directory on ibiblio but there are no packages in it yet.
    So, this is an example where a criticism has been posted suggesting that I have ignored a bug, whereas I have been working on it.

  23. pakt Says:

    “— anyone test foomatic?”

    Barry, the test I made on my Brother HL-2030 requires foomatic, so foomatic works too. Uses the same PPD file as your HL-2040 - no problem printing.

  24. RMP Says:

    Testing alpha2 and found a few issues.

    Mplayer is fine used embedded in the browser but is very prone to hanging everything when used standalone. Cannot connect to my wireless network using a USB dongle with the Zydas chipset and WPA. On a 24″ screen can only get 640×480 16, maybe xorg will be needed.

    Just a few thoughts.

    At the moment there are 2 of each of the following: Text Editors, Console Editors, File Finders, Batch File Renamers, cd Song Rippers. Is this duplication needed?

    I prefer all the Seamonkey suite. Don’t like DKOP or CDW, Grafburn is so much better. Don’t like Fotox, prefer GQview to sort and catalogue pictures and something like GIMP to work on them. I think Skype is needed but maybe it would have to be a PET. I know there are size constraints.

    The colour scheme is fine and the speed is absolutely great. Thank you for all the work and effort.


  25. davidchapman Says:

    I don’t have enough expertise to really contribute to debugging in Linux. But as a Windows developer - humble apologies - I recognise the eternal difference of approach between developer and user and I know how dispiriting it can be.

    So as an observer/user since around 2.10, all I can say is a big thanks to Barry for all your work and I hope you find the will to continue.

    Every version I’ve looked at has been an improvement on the previous one. If only we all could achieve that…

  26. cb88 Says:

    Barry, I rarely connect puppy to the net so i have never used petget online to download packages…..( i have on ocasion though). The bug is that when a package is clicked on to install it in rox or thunar(if set up correctly) that the package merely copies to / and is never decompressed.

    don’t packages decompress in /root/.packages/nameofpackage_dir ? when a package fails to decompress woulnt the dotpet be left in /root/.packages? if instead of the package decopressing and the package directory contents being copied to / perhaps only the .pet it self is copied to /

    I have experienced this just copy to /root/.packages behavior in the past but only with low ram.

    Is it it still occurs after modifying the space detection….

    I hope that gives you an idea of what is happening

  27. Dougal Says:

    Barry, if you’re going to recompile Xorg, you might want to also compile the Noveau drivers (http://nouveau.freedesktop.org/), in case people are interested.

  28. grumpywolfe Says:

    Also Barry I have a hp scanjet 3570c and it worked quit well with puppy 4.0 A2. When I try and use mplayer using the menu it locks the system up only the mouse works have to hit alt ctrl backspace and use xwin to get back to desk top but if I use the icon on the desktop no problem but only have window no menu

  29. BarryK Says:

    grumpywolfe, thanks, I fixed the menu — when you run MPlayer from the menu, it runs ‘gmplayer’, which can lock things up. On the otherhand, from the desktop icon it runs /usr/bin/mplayershell, which is a script I wrote that tries to start MPlayer properly. I have now fixed the menu entry so it also runs mplayershell.

    Regarding only one window appearing, you will probably see two mplayer entries in the taskbar. You can right-click and move the window onto the screen — as it is there, just off the screen. I have reported this to the JWM developer.

  30. BarryK Says:

    cb88, I’ve never had the problem you mention with PETget, which is why it has been off my radar. Are you running in RAM only, no pup_save? How much RAM in your PC?

  31. cb88 Says:

    512mb ram… happens in 2.14, 2.16, 3.00 and 3.01 that i am sure of…

    with pupsave but I don’t think that matters…could matter

    sometimes it works other times it doesn’t

  32. Gene Venable Says:

    You might want to check out the International Herald Tribune (www.iht.com), which has recently added a text to speech feature to each article. After you have selected the article, you can click on Listen to article.

    This didn’t work for me in what I gather was the latest Dingo, but did work in the latest Fire Hydrant, which I think is based on Puppy 301. It also worked fine in my Nokia N800, a Linux-based computer.

    Regular Mp3 files played for me in Dingo, however.

    The Herald Tribune uses something called Read Player to produce the text to speech.

Rollback to CUPS v1.1.23

November 26th, 2007

Back in Puppy 2.16 and 2.17 we used a CUPS package created by pakt ( see http://www.murga-linux.com/puppy/viewtopic.php?t=13466) which was based on binary packages from Slackware 11. I used this as the basis for a PET package, ‘cups_espgs_gutenprint-1.1.23-8.15-5.0.0-2.pet’, that you can find in the PET packages repository at ibiblio.org. My package had some significant size reduction.

This worked well. Puppy3 uses CUPS from Slackware 12, version 1.2.x and this is where things started to go downhill.

For Dingo I have not given up on CUPS, instead I am trying an experiment. I have rolled alpha2 back to CUPS v1.1.23, instead of v1.3.3 used in alpha1. This is probably not so good in the long term, as the shared library API changes and applications may require the extra functionality of a more recent CUPS. However, not to many apps link directly with the CUPS shared libraries, and those that do are likely to retain support for older versions of CUPS (ghostscript for example).

For the record, what applications are affected by the version of the CUPS shared libraries? I found out by running ‘createpuppy’ in Unleashed, leaving out the CUPS package and choosing to do a dependency check. It identified these packages link again the ‘libcups*’ libraries:

gutenprint, gtk, ghostscript, libgnomecups, libgnomeprint, and epdfview. Probably the qt4 package does to.

So, I recompiled everything in T2 with CUPS version 1.1.23 and updated the above-list of cups-version-sensitive PET packages and the ‘devx’ file.

So, Dingo alpha2, which is actually version 392, will have this rolled-back CUPS 1.1.23. Unfortunately I don’t have a printer to test (I’m away from home all week), all I was able to test was jcoder24’s print-to-pdf — which did work, except that there was no indication that the job had completed — maybe jcoder24 can take a look at that?

Maybe I missed a link or a file, maybe not. The web interface came up nicely, I chose a printer — actually I do have an old HP printer here, without any cartridges so it won’t actually print — and that much worked fine, no password required!

This is pretty much identical to the Slackware-11 based packages that pakt and later myself created, except for Gutenprint 5.0.1 instead of 5.0.0, so anything extra that might be needed ca most likely be grabbed out of those packages — in case anything needs to be fixed.

I invite you guys who know more about CUPS than me to play with Dingo alpha2 and knock CUPS into shape!

Note, I have listed gtk (and probably qt4) as linked with the CUPS libraries, but that is just a backend plugin and I will probably leave that out or make it optional in the future as I still want the option of using something other than CUPS. However, there are some apps — epdfview? — that absolutely must have CUPS to be able to print, which is a problem.

4 Responses to “Rollback to CUPS v1.1.23”

  1. disciple Says:

    “except that there was no indication that the job had completed ”

    Presumably you did have notification turned on?

  2. BarryK Says:

    disciple, what do you mean by “did you have notification turned on” …since I don’t know what that is, I probably didn’t!

  3. brucehohl Says:

    I booted Puppy-392 on my Compaq PC with USB HP Deskjet 5440 printer. Puppy booted without problem. I was able to set up the Deskjet 5440 without problem (no password requested). Printing from Abiword was good. The printer showed up in Gnumeric and SeaMonkey as well. CUPS also “saw” my other printer connected to my Ebox PC which is running Puppy 217. Looks good to me.

    By the way the new desktop theme caught the eye of my 9 year old daughter - she approves!

gtkmm-based applications

November 26th, 2007

Dingo alpha1 and Puppy 3.01 have Gparted (drive partition manager) compiled statically with the gtkmm libraries.

For Dingo Puppy alpha2 I have compiled all the gtkmm libraries, and Gparted and Inkscape are compiled to use these shared libraries. I have always been reluctant to do this as the gtkmm libraries are very big, almost as big as the gtk libraries that they provide an OO layer over.

So, alpha2 has the full Inkscape!

Actually, what drew my attention to the gtkmm libraries is I read an announcement that the Passepartout project is active again. This is a desktop publishing application, like Scribus except that it uses gtkmm. The project became dead a couple of years ago, but recently the author has revived it, and finally I was able to compile it — I had to get the very latest out of SVN, the 0.7.0 released version would not compile.

Passepartout would be a welcome addition to Puppy, and further justify inclusion of the gtkmm libraries, however when I tested it, it crashed or froze with just about everything I attempted to do. So, it needs work, but I can see the infrastructure is in place and I would like to encourage the author to keep at it.

Read more: http://www.stacken.kth.se/project/pptout/

2 Responses to “gtkmm-based applications”

  1. raffy Says:

    If added size is an issue, a smaller graphics program may be included, like Xpaint. Searching for it led me to http://www.linux-foundation.org/download/ - this could be a helpful list.

  2. razvan Says:


    Do you think its possible to provide us with a “pet” package for Inkscape 0.45 - this version comes with important features and it would be great if we could use it in our puppy linux.



Taking steps to prevent fragmentation

November 26th, 2007

I’m working on quite a few different things right now, and I’m looking at how this fragmentation of my time can be minimised. Some compromises are in order.

1. The kernel.
For Dingo Puppy 4.00 I’ll stay with the same kernel as used in Puppy3. Exactly as compiled in Puppy3, so all modules in Puppy3 will work in Dingo. The disadvantage of this is that modules cannot be compiled in Dingo due to the different gcc version. A reasonable compromise I think. I have to put support for the kernel on hold due to the error with generating security keys, that I have not been able to fix. Note, at some future date, I’ll move forward to a later kernel and dump the old ‘hd’ drive support as the scsi-layer support for ide pata drives seems to be maturing.

2. Customisations for Classmate (etc)
I’m not going to compile a special optimised kernel for the Classmate PC laptop (unless Puppy becomes super-popular in this hardware), nor will I build a custom Classmate-version of Puppy. Instead, for the Classmate and any other of these new cut-down laptops, I will just write an installer script. That is, there will only be one Puppy, and the Setup menu will have “Install Puppy to Classmate PC laptop” — you would have to boot Puppy on the Clasmate somehow, most likely by USB Flash drive, then just select this from the menu, and it will backup whatever is previously installed to the Flash drive (2GB drive required) then install Puppy and make any optimisations required.

2 Responses to “Taking steps to prevent fragmentation”

  1. RMP Says:

    I don’t think any of the new generation of small laptops needs a custom build, it would be better to go for the maximum hardware support possible for them in the standard puppy.

    They are so new it will need a bit of time to see how the current version runs on each of them. One blessing is the hardware should remain fixed in them so once compatibility is achieved maintaining it should not be too difficult. It is not exactly easy to get hold of any of them at the moment though!

    Maybe start a page on the forum for them and get the feedback as more people start using them and try puppy.

  2. kirk Says:

    “For Dingo Puppy 4.00 I’ll stay with the same ”

    That will be a problem for those who want ATI or Nvidia 3d drivers. Is there some rush to get Dingo out? It would be nice to have a kernel we can stick with for a while, especially since Dingo has all of the latest base packages. With those new packages, even a year from now, Dingo probably wouldn’t be that out of date. Take it easy Barry. Puppy should be fun, it is for me. Thanks for all your work.


Dingo alpha2, slight delay

November 24th, 2007

I was going to upload Dingo alpha2 today, but there will be a slight delay as I’m recompiling everything in T2. It takes about 24 hours, then I will need to modify some PET packages.

In T2, I had previously chosen the CUPS package and set that as the default print system. This causes some applications to compile with a dependency on CUPS, perhaps requiring the CUPS libraries. I’m not sure, but maybe GTK and Qt also have this dependency. I have now chosen the LPRng package instead and set that as the default print system — as my understanding is that this is just utility applications like ‘lpr’, no libraries that can become dependencies — meaning that any other print system that uses an ‘lpr’ utility can be substituted when Puppy is in use (like CUPS or PDQ).

Up to and including Puppy 2.17, all apps were generic, as the T2 build I did back then for the Puppy 2.1x series did not have CUPS as the default system. So, we had the flexibility of choosing later on what system we wanted to use. We had PDQ as default, but there was a CUPS PET package that could be installed. I like this flexibility, so that’s what I’m returning to.

I’m even thinking of developing my own Puppy Printing System — I think the acronym ‘PPS’ is taken! Maybe ‘PPQ’ or even ‘PUPS’. I got inspired when I looked at ‘lpr-bash’ — this is a simple Bash script that is a complete printing system. It is a script named ‘lpr’ and supports local printing and network printing using Samba. It would need to be enhanced with Gutenprint/HPijs IJS, ppd, foomatic support — not trivial exercises!

13 Responses to “Dingo alpha2, slight delay”

  1. Flash Says:


  2. lobster Says:

    “Pawprint” :) Good name Flash

  3. veronicathecow Says:

    Hi Barry, I’ve always been a hardware nerd as well as mucking around with Linux. I’m not sure if the 24 hours you quote is for a fully automated process but if it is and you have lots of ram..
    You could try setting up a very large Ramdrive. Load all I/P packages and progs into RAM and send all O/Ps there and you might find a significant difference in compiling speed.
    Hope it helps

  4. dvw86 Says:

    I have always had more luck with PDQ in Puppy than with CUPS. Although when CUPS does work it is very nice. I would not be opposed to going back to the way things were. My biggest concerns with PDQ would be things such as printing photos and using apps that want CUPS.

  5. JustGreg Says:

    I agree with Flash and Lobster. Pawprint fits! Oh by the way, PPS is a Russian acroynm for a class of submachine gun using 7.62×25 mm pistol round used in World War II.

  6. Sage Says:

    Barry, you need to get this one out the door PDQ. There’s a letter in the post from Kev. inviting you to become Minister for Computing, in charge of his OLPAC project…. Not, it’s not a typo; wherever you see an extra ‘A’ you can expect some action.

  7. mbutts Says:

    I like “Pawprint” as well. Since everything in Dingo is new it might be a good time to try a new print system. Especially early on then things are easy to change, add or remove. Puppy already works head and shoulders above some distro’s in certain areas and if printing was a snap as well that would push it ahead even further.

  8. jeffrey Says:

    lpr is a Unix standard, it’s simple, and it’s flexible. PDQ and CUPS are Linux systems (as far as I know). As such, if lpr’s filter scripts can be used to apply printer-specific control sequences based on some up-to-date repository of printer definition files, then it’s got to be a winner. The Web-based CUPS admin approach gives me the shudders. Why do I have to open a Web browser to restart my printer? It seems to be the wrong tool to me. (Similarly, I also can’t understand why major computer manufacturers are using Web-based interfaces to control remote servers. I find the Web-based approach rather hit and miss - sometimes it connects, and sometimes it doesn’t.)

  9. raffy Says:

    Gotcha! It’s good that jeffrey visited - he can help a lot in the subject of printing.

  10. brucehohl Says:

    Some (perhaps all) aspects of CUPS can be managed from the command line including listing jobs (lpstat), canceling jobs (cancel), starting, stopping, add, delete printers (lpadmin). The web tool might be viewed as optional.

  11. veronicathecow Says:

    Cups is the only thing in Linux that has ever got a printer working for me in Linux, printing is essential for take up by normal mortals like me(okay I’m a Windoze nerd)
    Reverting to the command line to do anything would be a big mistake if you want Puppy ro be used by your average user.
    Thanks for listening.

  12. Sage Says:

    Not necessarily. If exact, detailed but concise installation instructions are clearly displayed at the appropriate location, this would not be a deterrent to the installer or remasterer even if not fully understood by the end user. Barry is better than most with explanatory text, but will he be able to spare the time?
    This thread seems to have become entangled with the previous one?

  13. crafty Says:

    Well if one can do most aspects of CUPS also from the command line - then maybe one of our intrepid puppy coders could come up with a fairly simple GUI incorporating these ‘command line’ methods - would certainly be much easier than the web-based approach..



November 24th, 2007

As I have been rethinking all of the fundamentals for the new Dingo Puppy, perhaps I should also revisit the reasons for moving from PDQ to CUPS for printing.

What has caused me to raise this question is my current frustration with foomatic-rip — I cannot  get it to work with CUPS 1.3.3. When CUPS works, it’s really nice, when it doesn’t you’re in deep s***. CUPS is so darn complicated — ok, my knowledge of CUPS is very limited which doesn’t help when I’m trying to fix it. I was never able to fix that password thing (it asks for a username and password) despite much fiddling with the config file. I had to set certain files and devices with global write/execute permissions to get CUPS 1.2.x and 1.3.x to work. I’m just plain not happy with it.

In Puppy3 the CUPS PET package is 1.5MB, compared to the PDQ package at 0.1MB, so there is a big size penalty. Of course, in Puppy3 we had to go for CUPS due to the Slackware compatibility requirement. However, Dingo is free of any such constraints.

There is one pont to make here. We never did develop the full potential of PDQ. It works with foomatic which we never got into. In theory, PDQ can support all printers that CUPS can.

The PDQ gui uses GTK1, however I did read a comment once of a GTK2 port, though I’ve never seen it. Perhaps the source will compile with GTK2 or only need minor hacks?

I’m thinking of Dingo in terms of forging an independent path, not bowing down to what other distros are doing. After all, if Puppy/Dingo becomes very popular then it becomes a standard in itself and other will follow Puppy.

I welcome thoughts on this. In particular if you can recall your days of using PDQ and why you welcomed CUPS (or not).

20 Responses to “Why CUPS?”

  1. rodney Says:

    hi Barry,
    i have tried most releases of puppy from 0.9 onwards and was looking forward to getting cups in Puppy, I have to say it has disappointed me, for some reason it takes absolutly ages to print to the point that I begin to wonder if the machine has crashed.
    This is on either a k6 500mhz or a Pentium 400 incidentally I have the same problem using Vector 5.8 while Mepis3.4 and Vector 5.1print OK with cups on the same hardware. so perhaps its a kernal thing.
    This is such an issue for me that I had to go back to 2.16 from2.17 , 3.00 is no better.
    So for me a return to PDQ would be welcome.
    While I’m on the subject there also seems to be somethig amiss with the printing in Grafpup2.00 I get an incomprehensible screen, then it says i have chosen when I haven’t and offers a very limited choice of drivers, shame as I really like it otherwise.
    Sorry to sound negative I actually really like puppy and find it very useful.
    looking forward to Dingo, especially if you choose Opera.
    Thanks For your brilliant work.

  2. lobster Says:

    Save trees. Print on screen. :)

    PDQ worked for me. So does CUPS. Neither is straightforward to set up for the novice but printing never has been, or so it seems.
    From what I remember a lot of users were using Puppy as a ‘non-printer distro’ with PDQ. Which is fine for paperless office situations. CUPS made us a distro with much better real world printer support.
    It is possible that Puppy DINGO in Classmate can be provided with a recommended printer and/or supported printers that would make the PDQ option viable.
    If you control and can rely on the hardware, PDQ makes sense.
    CUPS makes better sense for a ’support the hardware available out there’ option.
    For me a computer that can not browse is not viable. For many; ‘no print?’ No thank you!
    CUPS can and has been provided as an essential add on. Puplets will arise if PDQ is not up to the job.

  3. Firefox Says:

    I think there is no printer driver in pdq for hp710c printer and I`m sure others are in the same boat.
    foomatic-rip and foomatic-gswrapper seem to go together (Perl scripts) in /usr/local/bin, made executable then linked from cups filter folder ln -s /usr/local/bin/foomatic-rip /usr/lib/cups/filter/foomatic-rip
    Hope this helps.

  4. PaulBx1 Says:

    PDQ standard, CUPS as a petget?

    I never got any of our small number of printers working with PDQ. With CUPS I got one working, the one with the plugged printer head. :-)

  5. Sage Says:

    PDQ was a doddle with all my old HP inkjet printers (the only make that reliably works after standing idle for six months!). Although CUPS usually works in most KDE distros (I don’t like Gnome), it has been a real pain in Puppy, failing in a couple of instances.

  6. tronkel Says:

    If both printer support and network printing in PDQ could be improved, I could live with PDQ. End result would be the same. CUPS as Petget should be fine just in case something goes wrong with PDQ

  7. tronkel Says:

    Also forgot to say, since Apple now controls CUPS, might be wise to use something else. Who knows what Apple might do with CUPS in the future as regards licensing etc.

  8. raffy Says:

    Many HP printers (especially the LaserJet 5/6 and 1200 series) have been easy to setup with PDQ. I used to print regularly to Win machines using the samba printer interface (the script is in the forum, with some PDQ problems troubleshot by jeffrey). I recall that I had no problem with PDQ in the Puppy 1.08+ series, but got into trouble in the 2.0+ series.

    Generally, I stayed away from CUPS, but welcomed your method of auto-generating the printer drivers, which keeps the installation small.

    I tend to welcome a PDF conversion utility than use CUPS for actual printing, but then PDF generation seems to require the same printer utilities.

  9. ferikenagy Says:

    I couldn’t make to run pdq with brother HL 1430 laser printer, but it works with cups + hl1250.ppd on puppy 301, as local and lan printer too…..it’s ok to return to pdq but shall it work?

  10. RMP Says:

    I have only used CUPS, it works, but its a bit of a pain if your printer is not listed and you have to download the printer description file, put it in the right location and go through the setup again.

    PDQ looks good and its small, but not tried it. If there was the ability to simply click on a button if your printer is not listed and you can go to an online list of additional printers to add would be great. Or maybe no printers in the basic setup and simply download the ones needed.

    I think the ability to print is essential and we are in a bit of a corner because the printing system that has support for the largest selection of printers probably has to win even if it is not really the best. I would be in favour of PDQ, and Apple now owning CUPS mmm.

    A bit off topic, but with the new generation of mini laptops such as the Classmate, OLPC and the Asus eee with fixed hardware it would be great if puppy could fully support each of these machines.

  11. pakt Says:

    Barry, if you can get PDQ + foomatic to print to your Brother HL-2040, then I’m all for abandoning CUPS and switching to PDQ :-)

  12. nic2109 Says:

    I too was happy with CUPS as I hadn’t got my network attached printer working for both printing and scanning - it’s an “all-in-one” printer/copier/scanner - under PDQ. This could well be ignorance plus failure to try hard enough but it was do-able (though not exactly “easy”) with CUPS as there is a .PET with the drivers and XSANE for scanning.

    I don’t mind which way this goes as long as it’s reasonably simple. I don’t insist on one-click installation - though it would be welcome if possible - but I did struggle a bit with CUPS where it asks for the protocol to use (i.e. IPP or http or socket etc) as I’m the sort of user who thinks that “localhost” refers to the landlord of my village pub. It was definitely a case of trial and error. I was also disappointed that attached via USB I can get 600 dpi which is optimal but when networked only 300. The difference is noticeable.

    Basically, as long as there are clear (and up-to-date) instructions it’s a case of “How Happy I’d be with Either”…… if they both can be made to work.

  13. jcoder24 Says:

    Since a lot of the printing functionality of CUPS is provided via ppd files, it should theoretically be possible to get most printers (that use PPD files) working with PDQ via a utility called ppdtopdq. ppdtopdq was included in the first pdf printer for puppy which was for PDQ. The extra functionality in the CUPS pdf printer should be easy to port to PDQ.

  14. Dougal Says:

    Did the people who had problems with CUPS ever use Pakt’s CUPS dotpet? Maybe it’s just a matter of setting it up properly.
    It would be a shame to just abandon it without trying properly, as it’s supposed to be pretty good and some apps (EpdfView) only support printing through it.

  15. Don G Says:

    CUPS is the only way I have ever been able to print from Puppy. I have a HP5610 All-inOne

  16. nic2109 Says:

    I’ve tried it both ways - that is both “native” from the CUPS website, and via Puppy’s Package Installer. Both worked for me.

  17. jeffrey Says:

    I could only get PDQ to print to my HP LaserJet 1100, not the Canon or Epson injets that I bought in hope, nor the Brother HL-2040 laser (which was most disappointing). I was very happy when Barry was given the same model and he immediately switched Puppy from PDQ to CUPS. CUPS’s Web-based interface is incredibly slow and non-positive (due to the lag in starting the browser), but apart from that it seems to work well. Pity it’s so big. I wonder if PDQ now supports more printers…

  18. brucehohl Says:

    One thing I really like about CUPS is that it has the ability to “broadcast” printers. So if you have a printer set up on one PC, all other PCs (with CUPS) can see and use that printer without any set up what so ever on the other printers. Thus, buy a printer, attach to a PC, set-up in CUPS, and now your printer administration is complete until you dispose of that printer. This is so sensibly simple you can see why Apple wanted this software.

    The web interface is also nice.

    I only used PDQ for one local printer so I don’t know if it includes the above capabilities.

  19. raffy Says:

    Puppy3’s CUPS PET package of 1.5MB should be worth putting to work. Bruce’s point above is a convincing argument. Also Dougal’s.

  20. wolf pup Says:

    the foomatic-rip, i was able to fix it by compiling hpijs from hplip http://www.puppylinux.ca/bugs that may help for hp printers.

Wireless “working” on Classmate

November 23rd, 2007

Thanks to pakt and tempestuous, I have fixed Puppy so that the rt73 driver loads instead of the rt2570 driver for the Intel Classmate laptop.

I’m using it right now, but I’m concerned about something… as soon as I brought up the wireless interface, the CPU became permanently busy, very busy. Also Blinky is showing a continual flow of incoming packets. Blinky’s light is contually on, and almost 700KB has come in just while I’ve been typing this …this definitely seems wrong! This concerns me, I’m going to disconnect now.

5 Responses to “Wireless “working” on Classmate”

  1. pakt Says:

    Barry, using top you’ll see a average ‘load’ of about 2.00 when running either rt2570 or rt73. This however is an illusion - the true load is *much* lower.

    In this thread


    forum member monkphunk wrote “ok, i think i figured out what is going on here. the interrupts are a red herring when it comes to the load. the rt73 processes are spending most of their time sitting in the uninterruptible sleep state, which linux includes in the load average. but they’re not really using much cpu time”

    I’ve gotten this result myself from both my RT2570 & RT73 USB devices. It’s a nuisance, but apparently harmless :-(

  2. PaulBx1 Says:

    Barry, when you say the CPU becomes permanently busy, do you mean that load indicator, lower right, next to the clock? (I forget its name.) I noticed that gives high readings as well, in 2.16, just copying a large file from the hard drive to a flash drive. Top says the CPU is at 1% but that load indicator says very busy. Makes me wonder what it actually is measuring.

  3. kirk Says:

    Yep, same on my rt2570 using the legacy driver. Don’t know if the new rt2×00 driver reports the same or if it’s even stable yet.

  4. BarryK Says:

    PaulBx1, it’s the ‘xload’ thing in the tray/taskbar where the red goes right up to two white bars and stays there.

    Okay, it’s an illusion, but what are all these packets continuously coming in from my modem/router? The “in” light on Blinky (the back screen in the icon) is green continuously.

    I need to get more info from Intel. Maybe there’s something funny with the hardware, as these laptops are designed to support mesh networking. The kids can bring their laptops together and I presume automatically all be talking to each other via wireless — though perhaps that’s all done in software.

    Yippee, requiring admin approval for the first post after registering has caught a spam comment!

  5. wolf pup Says:

    the incoming packets could be the beacons sent from the router to sync wifi devices. the wifi card would have to be in adhoc mode for it to share internet and files to other computers.

’sleep’ upgraded

November 22nd, 2007

’sleep’ is a Busybox applet, that has always been in Puppy. It accepts integer parameter in seconds. Busybox also has ‘usleep’ that accepts an integer in microseconds.

I was testing a script that I had downloaded, and it crashed. I investigated and found the line ’sleep 0.2s’ …hey what’s this? It turns out that the sleep utility in the ‘coreutils’ package accepts a floating point number, with units of s, m, h, d. It is also compatible with the normal integer-only sleep. It is only 12K and I just couldn’t resist replacing the Busybox applet.

One Response to “’sleep’ upgraded”

  1. ottod Says:

    Thank you. A week ago I was playing with aircrack-ng and ipwraw on Puppy and I found out some scripts did not work and the reason was the same you state here: a sleep function that just sleeps integer seconds, not floats. I had to fix the script but your patch fixes them all.

Humongous Dingo

November 22nd, 2007

Raffy will be pleased. I have fixed the Humongous Puppy. This has the pup_xxx.sfs and zdrv_xxx.sfs files inside the initrd.gz file, so there is only one file. Well, two actually, counting the Linux kernel vmlinuz.

In Puppy Unleashed, when running the ‘createpuppy’ script, you will need to answer certain questions as shown here, where a ‘no’ answer means just press the ENTER key only, ‘yes’ means press any character then ENTER. This is just the gist of the questions, not verbatim:

Do you want the modules inside the initrd? no
Do you want the modules inside the pup_xxx.sfs file, no zdrv? no
Build a smaller zdrv using pickmodules.sh? no or yes
Do you want a humongous initrd? yes

Actually, you could answer yes to the first question, but that still has some issues.

I created a humongous Dingo, just 75MB, and burnt it to CD. It works great, boots very fast.

Note, the reason for having a humongous Puppy is it is convenient for network booting if there is just one file, apart from the kernel. It requires a PC with at least 256MB of RAM.

4 Responses to “Humongous Dingo”

  1. raffy Says:

    Veeery pleased. Thanks, Barry.

    With this development, even student assistants can deploy cutting-edge Puppy in their school labs (See example lab assistants [Cherry Mae and Ailyn] at work here with Puppy: http://ec-tech.org/main/uc-2×2_typing-500.jpg ).

    Here’s a thread I started about this subject:
    I will update it when Dingo becomes available.

  2. raffy Says:

    Addition: rdesktop is quite handy in school labs if the users want to access certain Win services (IMHO a simpler option than the use of WINE or virtualization). I’ve recently tested version 1.4.1 and it’s able to do 16-bit color. I made JWM auto-hide the taskbar for better screen display (In this regard, it will be good if during the first use of rdesktop, a script will prompt the user about the benefit of auto-hiding the JWM taskbar).

  3. BarryK Says:

    Raffy. regarding your instructions in that forum link, the way I have built this latest humongous Puppy is with the zdrv file in the initrd.gz also, so it doesn’t have to be obtained separately.

    At first boot the zdrv will be taking up space in RAM, but after creating a pup_save file, it will be placed in the pup_save file at ‘/’, so won’t take up RAM space after bootup.

  4. raffy Says:

    Very nice, Barry. Thanks.

    I will make an explanation about Dingo and also make available a humongous initrd for 2.16 that contains zdrv_216.sfs.

Blog security increased

November 22nd, 2007

There has been some spam, so I have decided to up the security a couple of notches. Now, when someone registers, the first comment requires approval by me — all subsequent comments will not be moderated. Also, any comment with more than one hyperlink (URL) will require approval from me before it appears.

5 Responses to “Blog security increased”

  1. jeffrey Says:

    The security is a little constraining, unfortunately. I’ve finally managed to get WordPress to send me a password. I registered a week ago but nothing appeared, tried resetting the password, but still nothing. Tried again today and got through. There are probably others in the same boat. Pity that spammers are such a wasteful pestilence.

  2. BarryK Says:

    The confirmation email should reach you in about 10 minutes after registering. But, depending on how your email system works, it could go into your spam folder, or in certain circumstances even get bounced.

    The WordPress login window has a “I lost my password” link under it. You can click that to get another email. The confirmation email just has your password, it is not to confirm your registration as you would already be registered.

    If all else fails, send a personal message to me at the Puppy Forum and I can manually set your password and give that to you.

  3. tgeorg Says:

    Dear Barry,
    here is an article about the security of blogs with worldpress


    Puppy is the greatest LINUX. thanks

  4. raffy Says:

    Yes, jeffrey, it must be an email sending/receiving problem. I’ve had no similar problem as my Yahoo mail is used to exchanging mails with servage (the current host).

  5. uli Says:

    Hello to all, here’s my experience getting a password.

    Yesterday I have done the registration and did not get any mail from wordpress@puppylinux.com.

    After reading this article I tried the “I lost my password” link.
    After some minutes the first email was here,

    Subject: [Puppy Developer Blog] Password Reset
    Date: Thu, 22 Nov 2007 18:55:22 +0000
    From: WordPress

    just clicked the link and got errors

    Dont worry, 3 minutes later received the 2. mail

    Subject: [Puppy Developer Blog] Your new password
    Date: Thu, 22 Nov 2007 18:58:24 +0000
    From: WordPress

    hope it helps.

Classmate laptop!

November 21st, 2007

Intel has sent me two of their Classmate PC laptops, their answer to the OLPC. I just received them today, and booted Puppy from USB pen drive straight away! Hacao is already well down this path. Here is a snapshot of the Classmate alongside a normal laptop, running Hacao Linux:

classmate snapshot

Hacao Linux and the Classmate are discussed in the forum and in news reports:



Here’s an overview of the Classmate PC:


The two Classmates that I have received are a donation from Intel. What I plan to do is optimise Dingo for the Classmate hardware, including recompile the kernel. The objective will be even smaller size and extremely fast bootup, plus of course all the hardware must work.

13 Responses to “Classmate laptop!”

  1. tronkel Says:

    Puppy’s time has come. Why not supply the the Puppy OS as original equipment already pre-installed on a Classmate? OS would be free as always but good profit could be made on the hardware. Proceeds could be utilised for further Puppy development as well as provide a secure income source for Barry. This way he wouldn’t have to look for work in Perth and could therefore devote more time to Puppy/Dingo.

  2. lobster Says:

    Great News.
    Putting Puppy into hardware and optimising it makes perfect sense. I am hoping one day Intel might put Puppy into its CPU - Yep - “Puppy Inside” has a cute and familiar ring to it . . . :) Dingo. Bingo!

  3. raffy Says:

    With a core Puppy that you built, the Classmate (or similar education-oriented PCs) will have the best efficient OS for learning. Expect more low-power x86 devices in the coming months.

    Have fun!

  4. Sage Says:

    I know what folks will say (especially those who know my record!), but the old maxims pertain:
    Beware of strangers bearing gifts!
    No such thing as a free lunch!
    This would be an ideal opportunity to invite VIA, in particular, and a partner of AMD to submit samples……

  5. pakt Says:

    Great news, Barry!

    About time Puppy gets serious recognition :)

  6. No1 Says:

    Does anyone know how these will be priced for normal consumers?

    And maybe more specs like the resolution.

  7. mcgregor Says:

    Are you going to keep ‘normal’ Dingo, or it will be for Classmate PC only from now on?

  8. BarryK Says:

    mcgregor, of course not. A customised Puppy is just that, a special job. Probably what I’ll do is add it as a build option in Puppy Unleashed, a question “Do you want to build Puppy for the Classmate PC?”. Answering yes to this would build Puppy as per normal except for some substitutions, like a modified kernel and maybe some different packages. Likely this will be extended to other small laptops in the future — as Raffy says, there are more on the way.

  9. andrei Says:

    Given the slogan “one laptop per child”, maybe every Puppy enthusiast should get one :)

  10. BarryK Says:

    I’m running the Classmate right now! Booting off USB pen drive, I haven’t figured out how to install to the internal Flash memory yet.

    I’ve got the standard Xvesa X server working at 800×480, that was easy — you need to go into /usr/X11R7/bin/xwin and bypass the ‘ddcprobe’ and just run the ‘915resolution 38 800 480′ and set that as the only available resolution.

    Sound works. Wifi uses the RT2570 driver, but when I tried the Wizard and got to the place where I pressed the “Scan” button, it hung. So, lots to do, but mostly it’s running nicely …I have to get used to this tiny keyboard!

    I wonder if there is any way to configure the Symantec touchpad with the Xvesa server? I’ve never liked using a touchpad, but I’m trying to get competent with this one. Is it possible to do a “right click” by tapping on the touchpad?

  11. pakt Says:

    Barry, Puppy incorrectly detects the Classmate’s wifi as RT2570.

    In this thread


    tempestuous reports that you need to:

    rmmod rt2570
    modprobe rt73

    to get it to work.

  12. raffy Says:

    Barry, could you advise us about the best way to handle this trick in xwin? The discussion is here: http://murga-linux.com/puppy/viewtopic.php?t=24814


  13. site Says:



Gtkam, Sylpheed, Bogofilter, codecs

November 21st, 2007

Thanks to Rene and the guys on the T2-project for the patch that finally enabled me to compile Gtkam. The sticking point was the ‘libexif-gtk’ package, which is a dependency of Gtkam.

Gtkam is a GUI for libgphoto, for accessing digital cameras. Many cameras have a standard USB mass-storage interface so Puppy can access their photos. However, some do not, including the Fuji Finepix A610 that my daughter gave to me — beware manufacturer’s claims, as the Fuji online FAQ states that Finepix cameras have a standard USB mass storage interface, however in reality only some models do.
Unfortunately, libgphoto dos not have an exact driver for my camera, only a generic one, which only partly works. I can see the thumbnails but can’t copy the full images from the camera — trying to do so crashes Gtkam. Well, I’ll leave Gtkam in Dingo alpha2 and if you have a digital camera you can try it out. If it turns out to be too unstable or whatever, I’ll farm it out as a PET package.

I have upgraded SeaMonkey to version 1.1.6, and configured it without the Mail&News and Addressbook components. That leaves just the browser and the Composer HTML editor. Dingo now has Osmo personal organiser which has a nice addressbook, so I wanted to avoid the duplication of functionality. But more importantly this is the opportunity to save a lot of space. Everything in Mozilla is bloated, and by knocking off the Mail&News and Addressbook, replacing with Sylpheed, the live-CD goes down by almost 2MB. I also included Bogofilter, which provides advanced spam filtering for Sylpheed. It sure is nice to click on the ‘email’ icon and have Sylpheed appear virtually instantly.

On the subject of startup speed, I improved SeaMonkey by configuring with ‘flat’ chrome files rather than the more normal ‘jar’ files. The directory /usr/lib/seamonkey-1.1.6/chrome contains all the user-interface components, such as icons and themes, and this is a very big directory. Subdirectories in here are normally compressed into jar files, but this adds an overhead at startup. The chrome directory is compressed anyway as it is in the ‘pup_xxx.sfs’ squashfs file, so having jar files imposes two layers of compression. The ‘flat’ system leaves all subdirectories and files inside chrome as they are, just directories and files.

I’ve been testing Mplayer with various media files, in particular the Mplayer-plugin for SeaMonkey. These are some good links for testing browser plugins:

I have added some codecs to Dingo, but mostly I have been quite pleased with Mplayer’s handling of many media formats. I will be continuing to test today, in particular streaming audio.

I took out Pdrive v0.19 as it doesn’t work with my USB pen drives — lists it twice and ‘mount’ button just gives an error message. Back to Pmount.

17 Responses to “Gtkam, Sylpheed, Bogofilter, codecs”

  1. raffy Says:

    One way to make seamonkey the familiar browser interface to most people is the use of Liberation fonts. Cecc (from China) simply added them to TTF and reported good results. Dougal has added a forum page about fonts in Puppy:

  2. BarryK Says:

    Nice, I’m listening to BBC News Live, both the Windows Media and the Real Audio streaming audio choices work.

  3. BarryK Says:

    Ok, testing online video, I watched a BBC TV clip at this site:

    It offers Real Video or Windows Media, but the former plays sound only. I have had this problem elsewhere. Windows Media video plays fine, sound and video.

    I have tested the Flash 9 plugin, that works fine, sound and graphics. This site has lots of Flash:

  4. kirk Says:

    My wife has a Fuji F20, several months ago I built the latest Gtkam and libgphoto. It would crash when I tried copy the photos off the camera. Maybe it’s just a problem with Fuji? Nice camera though. Ended up buying a universal USB card reader for about 10 bucks.

  5. Pizzasgood Says:

    I recently bought an HP Photosmart E337. I haven’t tested on the latest Puppies, but I did have the crashing issue you mention in 2.14. Then I tried compiling the whole Gtkam/Gphoto thing myself, but Gtkam would still crash. BUT, when I tried copying them from the commandline with gphoto it worked fine. So it might not be the drivers that have the problem.

  6. Henry Says:

    My experience has been negative with gtkam, etc. Too complicated for what it does (and didn’t for me anyway), and doesn’t fit all cameras.

    As Kirk says, just get a card reader - I use a SanDisk Image Mate 5 in 1 USB reader. Yes, you do have to take the card out of the camera. Just use Rox or GqView to select thumbnails; edit with Gimp or whatever.

  7. Sage Says:

    Like I said previously, you’ll not be able to watch streaming BBC TV in anything except M$ apps until the New Year. This is because Tory Blare enjoyed an unhealthily close ‘relationship’ with Willy Gotes and rigged it with Auntie, amongst others. TB publicly admitted that he had difficulty finding the ON switch on a PC, but that the ‘cup-holder’ was a useful repository for his coffee in a photo-opportunity - I paraphrase. Don’t assume that Oz is the only nation with dud politicians - read the BBC News this morning! Bunch of Oxbridge Establishment chinless wonders - what’s your excuse? A retreat in the Outback may not be an adequate shield from their incompetence!

  8. disciple Says:

    Well, I signed up so I could suggest updating the link to this blog, but I was beaten to it :)

    Here’s an interesting quote from the GTK people:
    “GTK+ changed fairly substantially from version 1.2 to 2.0, much more so than from 1.0 to 1.2…
    Nonetheless, most programs written for 1.2 compile against 2.0 with few changes.”

    Maybe we should look at getting some old programs like LinNeighborhood working with GTK+2

  9. BarryK Says:

    Well, GTK 2.12 is a lot further down the track. I tried 8 - 9 old GTK1 apps, only managed to compile 2 or 3. Only one of them compiled as-is, that was Gxset, a couple of others needed minor hacking. The rest would probably compile with a lot more hacking of the source. The simpler one are easier, Linneighborhood is bigger and would probably be a major undertaking to compile. Anyone is welcome to try. Our new samba project is looking good though!

  10. fwiffo Says:

    I second the inclusion of the Liberation Fonts. I also discovered the midori web browser the other day. No gnome dependencies, just gtk and webkit as core.


    Alpha status yet, but interesting project. Osmo looks great, and to sylpheed and bogofilter I’m already accustomed.

    Cheers and keep up the good work.

  11. GreatnessGuru Says:

    SeaMonkey 1.1.7 candidate builds - please help testing!


  12. RMP Says:

    Been watching the development with interest and just a few thoughts.

    The change to Mplayer is a good move because in a head to head on a machine with the same codecs, it will always play a wider variety of media streams than any of the Xine family.

    The move to eliminate too much duplication of functionality in programs is also excellent. I hope with this new build from the bottom up we will have a set of programs that do their job and can be relied on for future releases.

    I am disappointed with the stripping out of the mail and news from Seamonkey. I use all of them and find them excellent. Size is certainly a big issue and I can see the reasons but the all in one internet suites like Seamonkey and Opera are excellent one stop shops. If these functions are left out of the final will it be possible to add them as pet packages.

    Thank you

  13. Dougal Says:

    RE: Liberation fonts: As I mentioned here in the past, at least the release when they were announced did not support non-Latin characters, so I don’t know how much “utf8″ they are.

    RE: Midori: it uses Webkit, which is the KDE (Konqueror) web engine, so there might be KDE dependencies.

  14. BarryK Says:

    Dougal, there are a few browsers under development that use webkit/gtkcore and I compiled a couple of them, not Midori though, and if I recall rightly there was no KDE dependency problem. Well, I haven’t looked closely at the Midori package, but presume it is using the gtkcore port of webkit.

    RMP, Sylpheed is integrated with the new SeaMonkey package, when you click on an email link. I’m not sure, haven’t used Sylpheed in a long time, but I think it also has its own addressbook. I can’t see any significant disadvantages of this combo.
    If you want SeaMonkey with the Mail&News, you would have to install the full SeaMonkey package, which will be a PET package.

  15. disciple Says:

    I meant to ask - without the mail and news, how does seamonkey compare in size/speed/whatever with firefox + some other html editor?

  16. PaulBx1 Says:

    Yes, eliminating Seamonkey email obviates the utility of Seamonkey browser, since half the point was the sharing of code. (Finally got a chance to use that word, hope I got it right :-)

    Possible flies in the ointment:
    1) can my current email messages be imported into Sylpheed?
    2) is there any hope of some day getting Enigmail to work with Sylpheed?

    I don’t have any use for Composer. Does anyone? My “composer” is Geany.

    I don’t think a cost of 2MB, out of something like 70MB, for a major application, is that bad. Perhaps memory consumption is a more important consideration?

    Seems like you might either keep both Seamonkey browser and Seamonkey email, or get rid of both and try Opera. Then all Seamonkey bloat is gone. Just a suggestion! I’m not an Opera fan, know almost nothing about it, but I’d give it a try. I wonder if it can import all my bookmarks.

  17. BarryK Says:

    A WYSIWYG HTML editor is one big gap in the Linux apps range. I used to use IBM Homepage Builder in Windows …I wonder if that would work in WINE …that’s the one thing I still miss from my Windows days.

    We have a choice between SeaMonkey Composer and Amaya. Amaya comes in two flavours, a smaller GTK version, and a bigger GTK+wxwidgets version. I have already been down the track of putting Opera + Amaya together, and counting the extra Qt libs it came to about the same size as the SeaMonkey suite. And that was using the smaller Amaya — the bigger version is really big.

    Both of them, Composer and Amaya, are very limited, fifth-rate choices for creating web pages. Probably a professional would go for something like Bluefish, but that doesn’t suite most Puppy users who are unfamiliar with any kind of HTML (and all the other related markup and scripting languages).
    Between the two of them, I would rather use Composer.

    The Sylpheed project has been going for many many years and has remained consistently active. They probably have all sorts of import/export and plugin thingys. You would have to check their homepage.

Pdrive new, Pbackup update

November 19th, 2007

Zigbert has been developing an enhanced Pmount, with an improved GUI and additional functionality. I have put v0.19 into Dingo. Read more:


Zigbert has updated Pbackup, a CD/DVD/HD backup tool, to v3.0.1. Read more:


3 Responses to “Pdrive new, Pbackup update”

  1. Henry Says:

    I just downloaded, installed, and tested Pdrive v0.19

    I have a DVD drive, a CD drive, 4 partitions on the hard drive, and 2 USB flash “drives.”

    In 3.01 the flash drives were listed twice and neither could be mounted.

    In 4.00 the flash drives were listed twice and could be mounted but with a confusing sequence of failure messages.

    Otherwise it worked. Both Pmount and MUT work correctly for me

  2. Henry Says:

    Just to be clear, I meant above that each flash drive was listed twice, showing 4 total.

  3. Henry Says:

    I see this has been fixed in v0.20.

    Very fine, Zigbert, nice extras. My only note now is to correct the spelling of deamon to daemon ;-)

Pwireless, Pnethood, Ripoff updated

November 18th, 2007

Plinej is developing a wireless network scanner. I have updated the package to v0.7.2. Read more here:


HairyWill is developing a Samba GUI drive mounter for Puppy. This may replace the Gfilemanager application that we have been experimenting with in Dingo alpha1. Read more here:


Ripoff is upgraded to v0.8.3. This is a simple CD song ripper. Read more:


Note, for people hanging out to play with Dingo, alpha1 is not recommended as there have been many bugfixes and improvements since then. I will upload alpha2 on Saturday — I’m constrained to wait until then as at home my satellite upload speed is pathetic and on Saturday I’ll be at a relatives place and will have broadband2 speed. In the meantime I’ll keep working on various improvements.

6 Responses to “Pwireless, Pnethood, Ripoff updated”

  1. crafty Says:

    Barry - just take your time..!!

    A side note - is there any way in this BLOG to have the screen ‘return’ to the main BLOG screen after one “logs” out of the submit comments section - at the moment it just drops you back to the ‘Login’ screen - so one has to do several clicks on the ‘back’ arrow to get back to the main screen..?


  2. BarryK Says:

    The login window has some links underneath it, one of which takes you to the main blog page.

  3. uli Says:

    Barry, you are using the default theme of wordpress.

    Would you like to get a better blog?

    Sidebar always on right side? to satisfy crafty
    Indicating where you are?
    No links to the active page? Jakob Nielsen would be happy
    Search and hilite results?

    I have some experience in designing and scripting in wordpress and would like to donate my 2p.

    (I am using ubuntu on hd and puppy2.17.1 on live-cd for demos and transports)

  4. BarryK Says:

    uli, yes, I just did the minimum to get it going. A bit later I will come back and see if it can be improved, and input from yourself would be welcome. I also plan to overhaul my entire puppylinux.com site, maybe use Joomla.

  5. hairywill Says:

    Looking at the volume widget in dingo has been irritating me so.
    I’ve put the -bg switch into absvolume so you can set the background colour even if the transparency fails. It also has a lot more visual shininess. The start command in .xinitrx now becomes
    [ -f /usr/bin/absvolume ] && absvolume -bg ‘#ffc100′ &

  6. BarryK Says:

    hairywill, thanks for that, it will be in alpha3.

Network Wizard bugfix

November 17th, 2007

It’s a minor bugfix. The script /usr/sbin/net-setup.sh in package ‘net_setup’ had a slight problem with some paths. It deleted some ‘dhcpcd-*.info, .cache, .pid files located in /etc/dhcpc, but this path applies to older versions of the ‘dhcpcd’ package. Recent versions set a path in /var/lib and /var/run. I modified the script to test if /etc/dhcpc exists, if not then use the alternate paths. Note, these .pid, .info, etc. files may need to be deleted to remove all traces of a previous run of dhcpcd on a particular network interface.

Fast-boot, full h.d. installation

November 16th, 2007

I just did a rough timing of bootup speed, with “fast boot” turned off and on. I introduced fast boot in an earlier post — it speeds bootup by minimizing probing and loading the same set of modules as were loaded prior to turning on fast boot.

My test PC has a 2.5MHz Celeron CPU, 512MB of RAM and a 160GB SATA hard drive.

Without fast-boot, the time from hitting the ENTER key at the GRUB menu to the desktop fully loaded is 22 seconds.

With fast-boot, that drops to 19 seconds. Or maybe 19.5, as I just held a wall clock in my hand while simultaneously watching the screen. A stop-watch would have been better.

But, the time for the kernel to load, that is, the time from hitting ENTER at the GRUB menu to the /etc/rc.d/rc.sysinit script start executing (there is no initial ramdisk), is 13 seconds. So that is the main time-hog and there is not much I can do about that. When control is passed to Puppy, a mere 6 seconds is required to fully load the desktop.

…I wonder if the kernel can be told to bypass much of its own probing? If I was to compile a kernel to work on a specific computer, then I’m sure that 13 seconds could be knocked way down.

17 Responses to “Fast-boot, full h.d. installation”

  1. Sage Says:

    Perhaps relevant is how your investigations are progressing with Machboot, TexasFlood and co?

  2. raffy Says:

    Machine-specific kernel compiling should now be practical with the increasing use of system-on-chips. The question will be who can/should do it.

  3. Headfound Says:

    kernel’s for specific configurations - isn’t that how apple get their os so solid? They only make it to run on their hardware.

  4. JustGreg Says:

    Personally, I will not probably use the fast boot. I change hardware too often. For a three second reduction in boot time, I do not think it is worth it. I would rather have the probing and correct modules installed. This is my opinion. One of nice features of Puppy is it runs on almost anything that I have tried. Making hardware specific version would eliminate that feature. For system with a hard disk installation, it would be reasonable. However, for USB key installation to be used with different systems, one will have problems.

    Enjoy life, Just Greg
    Live Well, Laugh Often, Love Much

  5. Me Says:

    Barry, Leftfield Question, When Any Plans 4 SMP &/ or Core Duo Pentium Kernel in Dingo or Slappy? Anyway Thanx 4 th’ Fun :) Distro

  6. Me Says:

    Providing there isn’t any significant OverHead or Penalty 4 Good Ol’ Socket 7 PCs! ;)

  7. nyu Says:


    Your test CPU is too slow to get fast boot. Please change it to 2.5GHz :-)

  8. linuxcbon Says:

    This question can be asked to the linux kernel people.
    Booting time is depending on your hardware and how the linux drivers do reckognize it.
    It is possible that the linux drivers have a bug or can be improved.
    It is needed to find out which drivers take so much time to start and why.

  9. Dougal Says:

    Barry, there were some people who reported that the “uncompressing linux” stage actually takes a very long time on some newer machines (with SATA drives).
    I guess it might have to do with the probing of various HW, but I was wondering how much time might go into the “uncompressing”: I assume “make bzImage” means it’s compressed, but I wonder if we need it — does it matter if the kernel takes up 2MB instead of 1MB on your HD? Am I getting something wrong here?

  10. BarryK Says:

    Dougal, yes, I am puzzled by these figures I’m getting, as I do recall getting faster times earlier on my laptop, which only has a 1.47GHz CPU, but does have a IDE HD. I’ll do another full HD install on my laptop and retest.

  11. Me Says:

    Barry, Leftfield Question, When Any Plans 4 SMP &/ or Core Duo Pentium Kernel in Dingo or Slappy? Anyway Thanx 4 th’ Fun :) Distro.
    Providing there isn’t any significant OverHead or Penalty 4 Good Ol’ Socket 7 PCs! ;)

  12. crafty Says:

    I’m not fully sure of this - BUT - SATA = ’serial access’ ; PATA (IDE) = ‘parallel access’ ; and internally to most main chips in a PC processing is performed in ‘parallel mode’ THUS speeding up code / programme manipulation..!!

    Now to perform ‘access’ to a SATA drive =
    “OUTPUT to drive” - convert parallel data/address to ’serial’ - buffer it - then send to external drive;

    “INPUT from drive” - send ’serial’ data from drive - buffer it - convert to parallel - deliver to processing..!!

    I believe that even with the newer chip technologies of today - ‘parallel’ data/address transfer WILL always be a small factor faster than ’serial transfer - due to the inherent “convert & buffer” steps involved with SATA drive systems..!!

    AND - I also understand that some drive manafacturers (not all mind you ) are actually using normal PATA (IDE) type disk drives [parallel transfer] for the data storage - fitting purpose designed ’serial-to-parallel/parallel-to-serial’ converter interface boards in place of the normal control electronics boards found on standard IDE drives - AND marketing these as SATA drives (serial access)..!!

    One can only but expect that with this extra ‘parallel-to-serial’ conversion step - access times will be some factor slower..!!

    Add this ’slower access’ to then having to ‘decompress/compress’ data to-&-from the SATA drives - surely there has to be some noticable delay..

    Maybe you need to ‘tweak’ the boot “init” script to be able to determine if the drive that Puppy is booting from is a PATA (IDE) type or a SATA type - and then direct the ongoing processing of the script dependent on what ‘type’ was found..!!

    Just thought I’d throw in my 2 c’s worth..


  13. Dougal Says:

    Crafty, it’s not related to the init script — the slow part is when the kernel looks around and detects your HW, before init is run.

    I think the slow machines were new desktops with SATA drives — maybe it has to do with the SATA controllers and different BIOS options for them? It’s probably worth booting them with loglevel=6 and seeing if there’s anything slowing down detection.
    Or maybe it’s just that those new boards have a lot of fancy peripherals that take longer to detect?

  14. Sage Says:

    Celeron is always bad news, even Sempr0n is better! But it does have the advantage of giving you plenty of time to offer tea/coffee to everyone in the neighbourhood while you wait for it to add 1 + 1. That’s digital 1, so to speak…

  15. crafty Says:

    Dougal - thanks for that clarification..!!! Strong evidence suggests that the delay with the SATA drives is related to the newer controllers - maybe we need to ensure that the ‘kernel’ used in Puppy has ALL relevant ‘config’ options enabled as appropriate to help speed up this ‘hardware detection’ - and yes, you probably are right in saying:

    [quote] those new boards have a lot of fancy peripherals that take longer to detect[/quote]..

    The “loglevel=6″ hint is well worth considering..!!


  16. No1 Says:


    I have USB sticks dedicated for my laptop and main desktop computer, as well as a general for use on problem machines, etc. It would be nice if we could tailor our puppy for a specific machine, maybe with some sort of recognition of the machine (even a crude HDD id would work) and warning if it was used on another machine or if hardware changed.

    Sometimes it’s easier to just have a USB stick instead of doing a HD install, but a faster boot time is also nice.

    An option not to compress some or all of the images might also be an idea for people with more storage than sense… err, CPU power :-)

  17. veronicathecow Says:

    Hi Barry (thanks for sorting PW problems).

    10% is a good start to speeding up boot times. I read somewhere about compiling for 386, 486, 586 etc and the results were slower when compiled for newer processors! Perhaps that has changed with new compilers etc?

    As part of Fastboot could the drivers that are used for that machine be put in an alternative file to zdrv_300.sfs (fast-boot_zdrv_300.sfs) so that on a fast boot only this driver package is loaded which should be faster and use less memory?

    Also the Pup_save file could be fixed in a the fast boot variablev and save searching lots of different drives/partitions?

    Finally there are a lot of comments in the scripts. Perhaps they could be left in for the Alphas and Betas but stripped out with some automatic program in the final? This would speed up execution and form a smaller file (Not sure how much difference?)

A “fix” for SeaMonkey and

November 15th, 2007

Just to clarify, I’m testing with Dingo, not Puppy 3.01, but there does appear to be a common core problem.

I mentioned in the previous post that the old Mozilla suite works and so did SeaMonkey after Mozilla had been installed — suggesting that installing Mozilla had overwritten one or more files that were the cause of the problem. Yes, I have confirmed it.

The only files that overlap are in /root/.mozilla/, and I narrowed it down to two files, ’secmod.db’ and ‘key3.db’.

Here is an explanation of these files:

An NSS certificate database consists of 3 files: cert8.db, key3.db and secmod.db. cert8.db stores certificates and Certificate Revocation Lists (CRLs), key3.db stores keys and secmod.db stores information about available pkcs#11 modules.

If two of them, secmod.db and key3.db are already created (from the Mozilla installation), then SeaMonkey works properly, I can login to my blog.

So, what’s this all about? Why on earth would the kernel version make a difference? Would it be ok to include those two files in the SeaMonkey package, or does that cause a security problem?

It’s interesting, the ‘cert8.db’ that SeaMonkey creates seems to be ok. It’s the other two I have to get from Mozilla. Anyone got any knowledge of this area?

Just to clarify a bit more. Running SeaMonkey with, all sorts of operations cause a crash. Even just selecting Edit–>Preferences in the menu. ’secmod.db’ gets generated, but not ‘key3.db’ or ‘cert8.db’ and there is a crash instead. If I copy ‘key3.db’ and ’secmod.db’ from Mozilla, then SeaMonkey generates ‘cert8.db’ ok.

On the otherhand, running SeaMonkey with the kernel, I choose Edit–>Preferences, and immediately those three .db files get created and all is well.

5 Responses to “A “fix” for SeaMonkey and”

  1. tronkel Says:

    the fsck on bootup takes a long time on my old PCs here. I went in to the init script in initrd.gz and commented out the appropriate 2 lines. Now boots in under 25 secs in Alpha 4. Any chance of user option to toggle the fsck routine?

  2. tronkel Says:

    I’ve just been playing around with various browsers in Alpha4.
    Built-in Seamonkey 1.1.5 definitely crashes. Also tried Seamonkey 1.1.6. - also crashes out unceremoniously. Firefox and Opera won’t even start. The little built-in Netsurf (or whatever) browser also collapses.

    Could this be to do with the fact that there is no Xorg in Alpha 4?
    Opera appears to be the best choice here because of space-saving considerations. Hope it can be made to work somehow

  3. JustGreg Says:

    I have been using Puppy 4.0 Dingo Alpha with SeaMonkey 1.1.5 and kernel without a problem. One thing that I normally do when I start using a new version of Puppy is to copy the .Mozilla directory from the last version of Puppy to the new version. This pass all my passwords, email, settings, and bookmarks as well as all the other files. I have all three .db files present, but they were created by Puppy 3.01 or the early version 2.17. I have been doing this since Puppy 2.0, I think. This may be a solution.

  4. tronkel Says:

    JustGreg said:

    “I have all three .db files present.”

    I’ve just copied all 3 db files from a 2.16 version over to Dingo. All versions of Seamonkey are now OK in Dingo. Goodie!

  5. Henry Says:

    FWIW I copied the three db files from 3.01 to 4alpha - no apparent effect on gpg.

    Then I copied the .mozilla folder - same.

    I doubt this observation is worth anything, but there you are.

SeaMonkey and the kernel

November 15th, 2007

SeaMonkey still crashes with kernel

Dingo alpha1 was released with two different kernels, and SeaMonkey works fine with the latter, but with the former if I try to make certain selections with the left mouse button, like login or download a file, SeaMonkey crashes. There is no helpful information apart from “segmentation fault”. The Internet connection is still functional and I can restart SeaMonkey and keep accessing the Internet. Others also reported this phenonemon.

I thought that it might have something to do with the underlying packages that might be “kernel version sensitive”. With the help of tempestuous I have recompiled everything I can thing of with the kernel running and its header files and source present.

I then built a new live-CD, but SeaMonkey still crashed. So, I built a new ‘devx’ file with the header files in it, then compiled the latest SeaMonkey v1.1.6. The headers are right, all supposed kernel-sensitive packages are right, but the latest SeaMonkey also crashes.

That makes me wonder about something…. I installed our old Mozilla v1.8 PET package, and guess what, it works fine (using it right now). I don’t have a SeaMonkey 1.0.x PET package handy, but I bet that would work also.
So, the SeaMonkey 1.1.x series is allergic to older kernels? Well, the 2.6.18 series is not that old, but kernels after that have some pretty fundamental rewiring, in particular with regard to timing and scheduling.

Actually, this old Mozilla 1.8 looks okay — it was the last of the Mozilla suite, before SeaMonkey came on the scene — I compiled it on 25 September 2005. It’s also considerably smaller than the latest SeaMonkey, so I might just use it as-is for the “retro” Dingo with kernel. Or, I could probably use SeaMonkey 1.0.x from Puppy 2.16.1.

I should be grumbling, as I messed around for the last 24 hours recompiling packages and trying to setup the Dingo so that SeaMonkey would not crash, only to find that SeaMonkey itself is the culprit. But, I’m just pleased that at least there is a solution, of sorts.

However, the jury is still out. After installing Mozilla, I ran SeaMonkey again and I was able to login to my blog …so, it seems that Mozilla has provided something that SeaMonkey needs — or more likely overwrote some SeaMonkey file. I’ll check this out some more.

Note, I also did some work on the structure of Unleashed, to improve building puppies with different kernel versions.

6 Responses to “SeaMonkey and the kernel”

  1. tronkel Says:

    I tried the latest Seamonkey V1.1.6 direct from the Seamonkey official site. This has never crashed with me. I always end up doing this and it appears to give no trouble. I’m using it with Puppy 3.01 retro though.
    So what is the difference between this version compiled at Seamonkey and the version as compiled by Barry? Library versions maybe? Perhaps a Slackware library causing bother?

  2. Sage Says:

    Dare I suggest that Opera might be a lot less trouble?!

  3. amadus Says:

    I noticed when I used the Flash plugin v.9 it crashed all the time. Not the case with v.7

  4. tempestuous Says:

    Barry, an interesting development: I said earlier that Seamonkey works fine in my installation of Dingo.
    Well my installation is full HD … but I just booted Dingo from liveCD on the same computer, and guess what, Seamonkey crashes!
    So there’s some difference between running a full installation and frugal.
    I tried overwriting the standard dhcpcd with alternative versions, no difference.

    Then I compared what modules were loaded in both cases, and found that there are a few extra modules loaded when running frugal:
    … but this doesn’t look significant to me.

    Is there any test that I can run to debug this?
    Can Seamonkey generate an error log?

  5. Henry Says:

    Barry, I’m so glad you’re taking a closer look at Seamonkey now. The forum is absolutely littered with my bug reports and calls for help on it through several versions. I still think you made a good choice in the Seamonkey suite, including an excellent browser and email and a so-so html editor. But it’s not enough to work with those basic functions - it has to work with some important extensions, like Enigmail, which my business depends on.

    I was practically a charter user user of Opera (you converted me from it in Puppy, although I still use it in windows). It’s a fine browser but it doesn’t handle GPG at all, has a weird mail system, and is still rejected by some sites. So overall I don’t share Sage’s enthusiasm for it in Puppy.

    Back to the problem - it seems Seamonkey works with Enigmail, but ONLY if I download version 1.1.2 or later directly from the Mozilla (Seamonkey) site. Currently I use Puppy 3.01 with Seamonkey 1.1.4 downloaded directly, which works very well. I am also testing 4.0alpha (not retro) with the included Seamonkey. As with all previous Seamonkeys provided as part of Puppy, it crashes immediately if I click anything under the OpenGPG menu.

    I also noted that in testing Gslapt in Puppy the Slackware package did not work with GPG. On the other hand the Seamonkey for GPG rpm package in PCLinuxOS installed easily and worked perfectly in PCLinuxOS (I didn’t try it in Puppy, although it might be possible with unrpm.)

    I really think it’s important to get this fixed, even if it adds a few more K/M to Puppy.

    Here’s to success!

  6. cb88 Says:

    In DSL 4.0 firefox 1.5 is used… I believe that is is compiled as a static binary would there be any advantage to trying that in puppy?

    I would much rather rol back the version and start using older software instead of supporting the continued development of “less than beta quality software”

    If program is “final” shouldn’t it just work regardless of wht you do to it?

    I realize that you do not follow that philosothy barry but what I don’t understand is why big projects that have a huge impact like mozilla don’t they just add “features” not stablility

    Sorry, but seeing the same old same old in bug reports kinda starts ticking you off doesn’t it…

Dict, Pdict

November 14th, 2007

I compiled the latest ‘dict’ utility from the ‘dictd’ package version 1.10.9. This provides access to online dictionaries and thesaurus’s.

I rewrote ‘PBdict’ as a Bash script (uses Gtkdialog for the GUI) and it is now called ‘Pdict’. This provides a nice frontend for dict.


November 13th, 2007

Well, I don’t have a network to test on, but I did a quick test of the search button. Running ‘gfilemanager’ from a teminal, when the search button is clicked, it does state at the bottom of the Gilemanager window that ‘nmblookup not found’. However in the terminal window I see that it is found. ‘nmblookup’ resides at /opt/samba/bin in alpha1. You guys reported that ’smblookup’ was missing, but it’s ‘nmblookup’ and it isn’t missing.

I created a little script named nmblookup to replace the actual program, and found out that ‘nmblookup’ is being called without any parameters, which appears to be the problem. So, is there something that has to be setup first, so that a parameter gets passed to nmblookup?

13 Responses to “Gfilemanager”

  1. Dougal Says:

    Oh, my, this Wordpress thing is a pain…

    Barry, the “News” link on your front-page is dead — you need to update all the different links to the blog (including news4, probably…).

  2. cb88 Says:

    why is it a pain? logged in and changed my password without a hitch…. have yet to see how fast it loads on dialup though

  3. BarryK Says:

    All of news4 got imported into this WordPress blog.

  4. BarryK Says:

    One thing that surprised me, considering how popular WordPress is — I tested with XP and IE 6.0, and the entire right-side insert is gone. That is, all that displays is the posts. Calendar,links, search, all missing.

  5. crafty Says:

    Still did not receive a ‘password’ via email for my 1st attempt to register - then found that it wouldn’t let me re-register (using the same email address) - so had to re-register using a different email address and “username”..

    Anyway now can post to this BLOG..

    This format for the blog looks OK - but do miss the old layout of the original blog - was so easy to read and also to find things….!!

    This ‘wordpress’ requires some little learning..


  6. hairywill Says:

    sorry about the duff report about smblookup
    I’ve hacked at the source for gfilemanager and managed to get it to scan and find my server but I haven’t got it to mount it yet.

    In callbacks.c there are are couple of checks to attempt to open the file nmblookup I’ve hardcoded the path and now the checks succeed. The checks could be removed completely because they have no relation to the systemz call to nmblookup.

    I also had to fix my /etc/hosts file so that the entry at the top for this machine was the real network address and not the loopback one, then I had to set the hostname “host puppypc”.

    It may be a couple of days before I can look at this again.

  7. hairywill Says:

    that should be
    hostname puppypc
    and I think it might have been set already

  8. raffy Says:

    Barry, about IE, even the puppylinux.org scripts give it little chance. I guess we’re seeing how far the Web developers have chosen to ignore IE.

  9. Dougal Says:

    Barry, I meant have a link to this blog from the /news4 page — so people are sure to know of its existence.

  10. hairywill Says:

    Barry, I have got gfilemanager working I have been using the 0.26 version from freshmeat. To be honest I’m not wildly impressed with its features as it works like an FTP program. It never mounts the share, files are only transferred individually using put/get like operations and the options for different file handlers (music/video/etc.) only apply to files on the local machine.

    The freshmeat details show that it was last updated March 2005. I think that Magnus has done a nice job with the interface I just think that the use case for the program is quite limited. This may be why it hasn’t been updated.

    I’m glad that I have taken the time to understand how much of it works as it will help with the mounter I am working on. Using nmblookup should allow it to scan machines that only do netbios over tcp/ip.

    I will send you my modifications to gfilemanger later.

  11. hairywill Says:

    Barry, here is gfilemanager with callbacks.c patched to remove the checks for nmblookup and with some extra debugging output. Its functionality is fairly limited as I explain above. I don’t yet know how a c program should search and access files on the path so I don’t understand why the checks are failing. I have commented my functional changes with “hairywill” if you wish to peruse the source.


    My research on /etc/hosts seems to indicate that puppy does not format this in the conventional way, at least not like the red hat 5 boxes we have at work. A typical entry reads something like:

    # Do not remove the following line, or various programs
    # that require network functionality will fail. localhost.localdomain localhost
    x.x.x.x machine.domain machine

    Fixing puppys hosts file into this format means that hostname -i now correctly returns the machines ip address, I think this would be a good thing to do regardless, though I’m not where the changes should be made, interface-up?

    If you wish to use gfilemanager then hostname -i must work properly.

    After a while of experimenting, my XP box was giving
    Domain=[KANGAROO] OS=[Windows 5.1] Server=[Windows 2000 LAN Manager]
    tree connect failed: NT_STATUS_OBJECT_PATH_NOT_FOUND
    This error was fixed by rebooting the server but gfilemanger has no handling for this error. I have no idea how common it is likely to be, or whether it was just caused by my heavy handed experimentation with smbclient.

    To correct my post above:
    Nmblookup cannot be used to scan for machines but gfilemanager includes a tcp/ip scan that I may reuse.

  12. BarryK Says:

    Will, can your other samba scanner project be used as a complete replacement for Gfilemanager? If so, or if it’s close enough, maybe we will just forget about Gfilemanager.

  13. hairywill Says:

    My original intention before dingo was for a mounter that was less complicated than linneighborhood. My intention is that it will satisfy all samba share mounting needs. It does for me already
    Obviously it is early days regarding promises of 100% reliable. As far as I’m aware MU is the only other person that has tried it (it worked).
    For the benefit of others the forum thread is here:

“Welcome” window, Pidgin fixed

November 13th, 2007

In Dingo, the “Welcome” HTML page does not come up automatically on first boot. Instead, a highlighted text message displays at top of screen, inviting to click-here for getting-started information. It goes away after 55 seconds if not clicked-on, replaced with the amount of free memory.

I tidied up the handling of this. If the user clicks on the welcome text string, the HTML welcome page starts up and the two barks are now synchronised with the opening of the page.

Note, the reason that I implemented it this way is for those people who often boot Puppy first-time and do not save a session, or often boot with “pfix=ram”. They won’t be bothered by the welcome HTML page, nor the two barks, and the welcome text string is unobtrusive and goes away on its own.

Note, if we could somehow auto-detect that the keyboard is US-layout, then we could do away with that dialog also, at least for US keyboards, and Puppy would just startup on first boot, desktop launched, no questions asked, ready-to-go.

Pidgin does not work in alpha1. I traced it down to some missing icons that are needed at startup. I perform deep surgery on some packages to get their size down, and Pidgin got the treatment, but I went a bit too far. I’ve fixed it and ‘du’ now shows the package occupies 4316KB, which makes me cringe. Anyway, I guess we have to live with it. This is version 2.2.1.

8 Responses to ““Welcome” window, Pidgin fixed”

  1. ferikenagy Says:

    As I knew keyboard layout cannot be detected , and maybe isn’t necessary at all. It could be much easier to presume everybody has US keyboard, start puppy without asking directly in X mode, and from there the user should select an specific keyboard layout (in similar way we select different Xvesa screen resolution, after puppy is launched in x mode) .unfortunately there is no little language(keyboard layout) applet written for JWM taskbar, and I haven’t the skill to write it.Changing keyboard layout on the fly (without restarting Xorg or Xvesa ) it is simple by executing command:
    setxkbmap -layout us #Us keyboard
    setxkbmap -layout hu #hungarian keyboard
    setxkbmap -layout ru #russian keyboard
    Question is if somebody have the skill and time to write keyboard taskbar applet for JWM? Here it is a link with tinny flags of the world for applet.

  2. ferikenagy Says:

    with tinny flags of the world for applet.

  3. ferikenagy Says:

    link: http://www.jfitz.com/flags/flags_png.zip with tinny flags of the world for applet.

  4. ferikenagy Says:

    another link with flag of the word: http://www.famfamfam.com/lab/icons/flags/famfamfam_flag_icons.zip

  5. amadus Says:

    “PKEYS=fr” at the kernel boot prompt worked nicely for me with each fresh boot, to skip the kb dialog once for all.

  6. Sage Says:

    That’ll be tiny, feri - tinny has an entirely different connotation downunder!
    Pictures take up more bytes that text - stick with text, Barry.
    Assume a UK k/b, though - our Chinese friends prefer their English tutors to be, errrr, English and the entire sub-continent speaks better English than the British. Spanish, is, of course, the only important language of the Americas!
    Glad you rid us of the distracting dialogue.

    PS. ‘dialogue’, incidentally is the correct spelling, just that the wrong spell checker seems to be loaded …..

  7. cb88 Says:

    Sage why would we want to cater to people that are sending us lead paint on toys antifreeze in toothpaste and hard drives with preinstalled viruses?

    anyway I have nothing against chinese people just thier government…

    I mean really what kind of justification for changing to a uk keyboard is that anyway?

    wonder why u/k keaboards would even be different do ya’ll UK people know about more letters than we american HicKs do (american city slickers please don’t take offence ;) )?

  8. Pizzasgood Says:

    Slight differences, to take care of things like alternate currency symbols, a heavier usage of funky accented characters, and a couple minor positioning and shape changes.

    Personally, I wouldn’t even care if Barry had it default to Programmer’s Dvorak. It’s not like it’s that hard to change it. Actually, I should consider learning that one. It might be annoying when I want to install a game though, but I don’t do that much and it wouldn’t be too much hassle to remap them in the game or just temporarily revert to US keys.

Agave americana

November 12th, 2007

This is growing on my boundary fence:

agave snap

It’s known as the Century Plant, or Agave americana, and even has it’s own page in the wikipedia:


It was a young plant when I moved here in December 1998, so I reckon it would be at least ten years old. The soil is extremely poor, very arid, and it does well here. I’ve gots lots of others growing, including a variegated variety. Last week I glanced outside from my kitchen window and there was this enormous stalk — it must have shot up very fast. I’ll post another snapshot when it flowers!

7 Responses to “Agave americana”

  1. mrperson Says:

    Test comment, posted by mrperson (BarryK really), normal user.

  2. Bahurim Says:

    Nice plant! Reading wikipedia they say the plant dies after flowering. Will be interesting to see if that is true. You can also make mezcal from it, hmmm.

  3. Sage Says:

    Glad to report that my PW was directed to my Inbox, not the Junk folder.
    The Agave family have interesting medical properties - forget the details but sure they’re on the Net. Amazingly, quite a few grow in the UK, along with the Mahoniae from the California desert.

  4. caneri Says:

    yup..it got spammed (login)…agave is what they use to make tequilla

  5. cb88 Says:

    yeah we had one of those in a flower pot but it died I am pretty sure it needed to be in the ground…

    I haven’t seen a century plant bloom before but yucca plants have beutiful white blossoms. Yucca plants are notoriously hard to kill even if you burn them… if there is one little piece of root left you get a brand new one :roll:

  6. ilanrab Says:

    In my last house, in Los Gatos, CA I had one just like that. For 7 years it didn’t do a thing. Then one year it sprouted up like this one. Had beautiful
    flowers on top and LOTS of seeds — watch out!!! It was a huge plant,
    and it did not die.

  7. Flash Says:

    There were two large and deadly-looking agaves beside my house when I bought it 20 years ago. One bloomed last year, then died. Unfortunately the stalk broke before the flowers actually bloomed. I left it alone and it still bloomed, but I guess whatever pollinates the flowers (I think it might be bats here in Arizona) didn’t see these because they were too near the ground. Anyway, there was nothing like fruit that I could see.

    As for using the sap for anything medicinal, the sap of these particular agaves affects me like poison ivy. Wherever I get it on me I get a rash that lasts for over a week and itches almost unbearably. The first time it happened I wasn’t sure which plant was responsible, so I had to endure it twice to be sure.

Welcome to my shiny new Blog!

November 11th, 2007

After much suffering on the old blog, I have moved to WordPress. Also a new host, Servage.net.

Servage.net is now the host for my domain puppylinux.com, and soon I will move my old goosee.com here too. Note, we have another account on Servage that hosts puppylinux.org. There may be capacity to host everything in the one account, although we did have some overuse problems with puppylinux.org awhile back.

My old Blog still exists here:


The old Blog is read-only, so you can’t post.

Please note: I have setup WordPress so that you have to register to post, with email verification. An email will be sent to you with a password, but please be watchful when you read your email — I did a test registration as “mrperson” and gave my gmail email address, and after logging into gmail I found the email — in the “Spam” folder!


Barry Kauler

7 Responses to “Welcome to my shiny new Blog!”

  1. nic2109 Says:

    Hi; and welcome to Wordpress. I cannot have been the only person to have suggested it so I’m glad you are trying it out. It is just SO impressive how much you can do with it - I’m awestruck with the built-in capability.

    If you download XAMPP you can run it yourself on your own PC. It all works just fine under Puppy 3 (and maybe earlier versions too, but I had a strange permissions issue with 2.17), you’ll just need to chmod 777 /tmp so that mySQL can create temp tables and away you go.

    Thanks for all you do. Puppy AND botany lessons!

  2. billstclair Says:

    I resisted a MySQL-backed blog for a long time. I’m using Drupal, but Wordpress is also a good choice. Sorta like Honda or Toyota. A matter of taste.

  3. nic2109 Says:

    Call me naive (many do) but what’s this mySQL hang-up all about? For Wordpress all you need to do is (1) create an empty database; (2) edit wp-config with the db name, userid and password so wp can connect to it; and (3) remember to back up your database regularly. And if you are being hosted then the host should do (3) for you. You don’t ever, ever have to write an SQL statement, ever. What’s the problem?

  4. BarryK Says:

    I had problems sometime ago, on Servage and elsewhere, the mySql database kept getting corrupted and I had to go into the admin and type in some arcane command to repair it. I got turned off mySql then, but I’m prepared to give it another go.

  5. raffy Says:

    Hope you took time to activate anti-spam in the Options.

    As a result of security bullet-proofing, PHP code is often updated in blogs like this. Must be done regularly, but is painless anyway.

  6. raffy Says:

    … ah, that should be in Plugins (Akismet for anti-spam).

  7. BarryK Says:

    One interesting option is admin approval for the first post. So, after you have registered, your first comment requires approval, after that you are free to comment. I have that one turned off, but if spammers do take the trouble to register, then I’ll turn it on.

Plugins for SeaMonkey

November 11th, 2007

I have just noticed that SeaMonkey has no plugins in Dingo …but they are there! Dingo has a huge swag of them, residing in /usr/lib/mozilla/plugins. Previously I had /usr/lib/mozilla as a symlink to the seamonkey directory, however it is better for it not to be a symlink. This is the place for all plugins that any browser can access.

The fix is to tell SeaMonkey or Firefox where the plugins are:

export MOZ_PLUGIN_PATH=”/usr/lib/mozilla/plugins:/usr/lib/seamonkey/plugins”

Actually, I’m surprised that the SeaMonkey startup script doesn’t find them anyway.

I have put the above variable into /etc/profile, so it will be fixed for alpha2.

PupDial improved

November 11th, 2007

There were some suggestions from users that I have implemented.

When attempting to dialout, if it fails and the user has to hit the ‘abort’ button, PupDial now returns to the main window and retains the log of the attempted dialout in the ‘Connection status log’ box, so you can study what went wrong.

The password entry boxes no longer displays the actual passwords, a string of x’s instead.

2 Responses to “PupDial improved”

  1. BarryK Says:

    Friday, November 9, 2007, 10:56 PM
    well barry I had writen a nice long comment on here so i’ll just recap..

    4.00ahpa….memory hog…. I think it is GTK memory leak?

    was cacheing 400+ mb with most of the apps running but puppy extraced wouldn’t even be half that size….

    hardinfo was reporting 180mb actaul ram useage so that means that 128mb PC’s will be crippled if running from ram…at least i thing kinda hard to test for that if you have 512mb…

    I like the new gtk theme chooser…

    liked fotox sad to see it go please maintain dotpet…seems very functional but haven’t tried it on old lappy yet…

    install to usb doesn’t seem to work there was an error message….I forget it but it dissplayed with the installation was in progess….

    also is it possible to integrate fwe (frugal windowing environment)?
    would it still be possible to run X directly from fwe?
    if so it could replace the console with a window managed console
    fwe is based on fbui except that it is a kernel independent module….
    here is a link: http://home.comcast.net/~fbui/fwe.html

    Saturday, November 10, 2007, 09:54 AM
    barry I just remember one more thing…

    some ltmodems are not detected by pupdial…

    need to load ltmodem and ltserial first before probing

  2. nic2109 Says:

    Regarding the possible memory leaks …. there is a utility/package called “Purify” that can check for memory leaks and so forth in C/C++ programs.

    Ah; I’ve just Googled for it and it’s sold by IBM - note “sold” - so it’s probably not on the radar. If you think it’s worth persuing then there may be a gnu equivalent out there.

    If you want to see what IBM say about it then here’s the link :-


GImageView, GWhere, nmblookup?

November 11th, 2007

GImageView is a very nice GTK2 image file manager. You can do the usual stuff like looking at thumbnails of images in a directory and a slideshow, but what I really liked is it uses MPlayer to create thumbnails from frames of video files — as well as thumbnails of vidoes, it even plays them.
The only trick to using GImageView is that you need to double-click on things, which is not quite apparent when you first fire it up. It even plays videos full-screen. Home page:

I have taken out Fotox in favour of GImageView. The former has a bug anyway — the fill-window zoom button does not work. Interesting, there was another graphics app I tested that had the same problem, so I think it’s a GTK 2.12 problem.

Dingo alpha1 lacked a CD/DVD cataloger, as the old Gtkcat uses GTK1 and I couldn’t find anything suitable that uses GTK2. Jason (plinej in the forum) came to the rescue. He found ‘GWhere’. It’s really nice. It will also be in alpha2. Home page:

In the forum feedback for alpha1, I think it was mentioned that Gfilemanager needs ‘nmblookup’? That’s a utility in the Samba package, 552KB so I will only put it in if it’s really needed.

2 Responses to “GImageView, GWhere, nmblookup?”

  1. BarryK Says:

    Sunday, November 11, 2007, 03:22 AM
    I thought you said Gfilemanager was to replace LinNeighborhood? Is it intended as a second file manager as well?
    LinNeighborhood doesn’t work without nmblookup.
    I was just playing with Samba today, seeing how many things aren’t really needed, and it seems that you need nmblookup, smbmount, smbumount, smbmnt and I think smbclient for a Samba client using LinNeighborhood, need smbspool for samba network printing, and for a Samba server only need smbd and nmbd.

  2. BarryK Says:

    Note, I’ve restored Fotox for alpha2. I decided that Gimageviewer overlaps the functionality of Mplayer and ROX-Filer, so I can’t really justify its extra size. I’ll contact the developers of Fotox and see if the bug can get fixed.

    Puppy, including Dingo, already has ‘nmblookup’. What Gfilemanager wants is ’smblookup’, and we are discussing on the forum how to provide this — it may be just a script that calls nmblookup.

Firewall fixed

November 11th, 2007

The firewall is broken in alpha1, now fixed.

There were some bugs with module loading. One problem was that the firewall looks for module ‘ipt_state.ko’, which does not exist. Actually, it is superceded by ‘xt_state.ko’.

The pup_xxx.sfs file is supposed to have a basic set of modules in it, so they do not have to be fetched from the ‘zdrv’ file, but this basic set is missing from alpha1, due to a bug in the ‘createpuppy’ build script. The basic set includes all the modules needed by the firewall. I fixed the bug in the ‘createpuppy’ script, also a bug in the ‘init’ script that prevented the modules from being copied across from the initramfs before the switch_root when PUPMODE=5.

One Response to “Firewall fixed”

  1. BarryK Says:

    Thursday, November 8, 2007, 06:57 PM
    Is the latest elinks a possibility for Dingo (or Puppy)?

    Thursday, November 8, 2007, 07:37 PM

    Did you use Puppy as a build environment for the T2 scripts? If so, which version?
    Are you going to post a puppy-T2 buld package as you did for 2.10?



    Thursday, November 8, 2007, 08:30 PM
    thanks Barry

    Friday, November 9, 2007, 05:03 AM
    Yes, T2 was used to build Dingo. Yes, I’ll be making it available. The version of T2 I used is from ‘trunk’, which is their ‘8.0′ development version.

    Friday, November 9, 2007, 05:06 AM
    Oh yes, I did use Puppy as the build environment. However v3.01 has a couple of utilities missing. After adding those, it worked.
    Dingo has those utilities so will work for building T2.
    I can’t recall exactly… ‘flock’ was one of them.

    Friday, November 9, 2007, 05:54 AM
    Barry, a new version of RipOff was released yesterday.

    Friday, November 9, 2007, 06:28 AM
    Could the fix for searching and mounting pup_301.sfs and zdrv_301.sfs in RAMdisk be also included in init? Just like in the implementation of humongous initrd in 2.16. Thanks a lot.

sGmixer, AbsVolume

November 11th, 2007

There is a problem with Zmixer, at least on my laptop — the master volume control does not work. It cannot be dragged by the mouse, which is odd, as some of the other level controls do work. Anyway, I have substituted ’sGmixer’, which is another GTK2 application. sGmixer has one peculiar feature in its user interface, but a trivial thing that I can overlook for now — at least the level sliders all slide.

I would like to thank hairywill for turning AbsVolume into a true tray applet. It’s real nice. A left-click brings up the master-volume slider, a right-click brings up the full mixer. Thanks for the time and effort that you have put into that Will.

This leads me to something else. Dingo has a choice of three different full GUI mixers, alsamixer (ncurses), zmixer and sgmixer. There is also setvol, which is a console application. Alsamixer and setvol are always there in every Puppy, but for a GTK2 GUI mixer chose zmixer in alpha1 and it will be sgmixer in alpha2.
So, there does need to be a /usr/local/bin/defaultaudiomixer, which I have now created.

I have modified Will’s absvolume applet code so that a right-click executes ‘defaultaudiomixer’. To get the volume applet to startup, I followed hairywill’s instructions and put a line into ~/.xinitrc and modified .jwmrc.
…or rather, modified the template file in the jwm package, /etc/xdg/templates/_root_.jwmrc.

Note, there is one thing that we will have to put on the to-do list. AbsVolume displays the icon with a non-transparent background color, that does not match the tray background. I see in the code that the xpm image has a transparent background, but there must be some other setting in GTK to either set background as transparent or set to some specific background color (as in Blinky).

One Response to “sGmixer, AbsVolume”

  1. BarryK Says:

    Thursday, November 8, 2007, 09:02 AM
    I noticed the transparency problem after installing it into dingo yesterday. I have searched for a proper solution to using transparency, though not found anything. If nothing turns up I can implement a switch such as used by blinky. It is not a particularly good solution though as .jwmrc-tray will need to be fixed each time the gtk theme is changed.

    I haven’t tried installing the normal absvolume and using alltray, though this would add extra weight.

    You asked last week about a GUI for adding quick launch buttons on the tray. The JWM Configuration Manager already has an option to configure the panel buttons, though this doesn’t work properly because of the buttons that are commented out.


    Administrator (BarryK)
    Thursday, November 8, 2007, 05:50 PM
    Ah, so I need to get rid of those commented-out lines in .jwmrc-tray. Ok.

    Well, there is a kind of hack that could be used to set the mini-volume icon background. Instead of an internal xpm image, read from /usr/local/lib/X11/mini-icons/mini-volume.xpm and replace the “None” with the background colour before using it, without changing the original file of course.
    The background colour could be determined from the .jwmrc file …you would have to read the tag in the
    section I think.
    That would be an automatic system, if it can be done.
    Administrator (BarryK)
    Thursday, November 8, 2007, 05:53 PM
    It would also need to test the existence of ~/.jwmrc, so the applet will still run with other window managers.

    Saturday, November 10, 2007, 10:43 AM
    The absvolume transparency is not a problem. I spent a while considering the aggravation of parsing the jwm config in C amongst other options. Luckily before I did anything I realised that restarting X makes the transparency work.

    Here is a much shinier version:
    The slider now fills the window properly
    I’ve used the png files rarsa(?) included with mini-volume.tcl to generate a set of internal xpms that show different volume levels in the icon.
    I’ve also included the volume level in the tooltip.
    Both the icon and tooltip are currently updated once a second and also on every slider value changed event. Using the events this way may be a bit costly in terms of cpu but it looks pretty seeing the icon change as you move the slider.
    The file is now up to 20K stripped I will play at reducing the number of colours in the icons, I haven’t tested it yet but it may be that the transparency works because I’m now using greater than 16 colours.
    If you have an opinion either way I could use external images which would allow themeing, though it seems quite convenient to keep everything in one file.

    I think I understand more about why the skip_taskbar_hint doesn’t work there is a lot of casting between gtkwidget and gtkwindow going on which may well be the problem.

    I tested it in icewm and it won’t start in .xinitrc so I put it into /root/Startup (straight into delayedrun would be just as good).

    I think pushing the idea of $HOME/Startup more would be good though possibly making it hidden would be sensible. It makes a good place for users to turn things on that are not essential such as glipper and some of ziggys recent additions to .xinitrc


    Saturday, November 10, 2007, 02:19 PM
    I’ve been developing in 3.01 and transparency works when restarting X.
    Now testing in dingo I realise that it doesn’t. I wonder what the differenceis?


PPPOE without Roaring Penguin package

November 11th, 2007

Dingo has the Roaring Penguin PPPOE package but not the Tcl/Tk GUI. However, the ‘ppp’ package has a plugin, /usr/lib/pppd/2.4.4/rp-pppoe.so, that is code taken from the Roaring Penguin package, and it is possible to do PPPOE with just the ppp package, no Roaring Penguin package required at all.

The LinuxFromScratch people have been into this. They have a page that outlines it:

They also have a detailed mail-list thread:
http://linuxfromscratch.org/pipermail/b … 60173.html

The thread explains everything. The kernel used in Puppy 3.xx and Dingo alpha1 has the required modules, as per these config settings:

CONFIG_UNIX98_PTYS=y #this is builtin.

If we can get PupDial to do PPPOE with the ppp package, then there will no longer be any need to keep the Roaring Penguin package. It does not need WvDial either.

This would be a very interesting project for someone into using PPPOE! You don’t have to be using Dingo, this would apply to any of the earlier puppies — well, I would like to expand PupDial to handle this, and PupDial was introduced in Puppy 3.00. If we get this working from a terminal and the steps standardised, then I could translate that into the GUI.

Here are more howtos:
http://norean.freecontrib.org/nakooki/S … OW-TO.html

Note, this mode of connecting to PPPOE is called “kernel mode PPPOE”.

CORRECTION: I didn’t put the roaring penguin package in Dingo alpha1.

Asunder, PBcdripper, RipOff

November 11th, 2007

I don’t know much about CD song ripping, but I had a play with these three. Dingo alpha1 has the first two, and I have just compiled RipOff.

For simple intuitive taking songs off a CD, for the novice, I liked RipOff. It matches well with many of the existing audio libraries currently in Dingo.

Asunder is much the same, except tried to automatically do a CDDB download then the program crashed as there was no Internet connection. This can be turned on and off in the preferences. RipOff on the other hand just had a button that you click to get the CDDB info. Asunder automatically ticked all songs to be ripped, and unticking had to be done manually, whereas RipOff has ‘Deselect all’ and ‘Select all’ menu entries. Asunder can create an M3U playlist file.

PBcdripper 2.6 has a lot more options, or so it seems from all the things to select on the main window. After choosing ‘mp3′ encoding, it wasn’t entirely inutuitive to me that I also had to click the ‘Choose’ button to actually select mp3 encoding …seemed like an unnecessary step. But, I quickly got the hang of it.

All three are quite small, so I’m inclined to leave two of them in the alpha2 release, as each has a flavour that will appeal to different users. All support wav, mp3 and ogg, while PBcderipper also supports wma encoding. PBcdripper also supports id3tags. I think PBcdripper and RipOff are good choices, the former as it will appeal to the ripping connoisseur, the latter to the novice.

One Response to “Asunder, PBcdripper, RipOff”

  1. BarryK Says:

    Wednesday, November 7, 2007, 11:39 AM
    Hi Barry, do any of them do FLAC? This is a free lossless encoder that compresses files as well.

    Wednesday, November 7, 2007, 02:14 PM
    Asunder does, have used any of the others. Also the gpRip made Nathan (of Grafpup) also does flac. http://www.murga-linux.com/puppy/viewtopic.php?t=18592

    Wednesday, November 7, 2007, 02:15 PM
    meant to say I have not used the others.

    Wednesday, November 7, 2007, 02:17 PM
    Tony: PBcdripper can encode to flac if you have flac installed (I think, at least).
    The other two can, but have to be compiled with flac support — I think it comes in the form of plugins, so maybe Barry should compile with support for it.

    Barry: good, you found RipOff. I compiled it ages ago, along with Asunder, and since you mentioned Asunder I’ve been trying to remember what it’s called…

    Wednesday, November 7, 2007, 03:06 PM
    I’ve been very pleased with Asunder, it does flac.

    Wednesday, November 7, 2007, 05:55 PM
    PBcdripper searches your executable path for various encoding libraries, then presents encode buttons for the libraries found. These include -

    The faac library is for AAC/m4a encodings, particularly useful for iPod owners since this is Apple’s preferred codec.

    Wednesday, November 7, 2007, 07:51 PM
    My (quite off-t) question is whether rippers are must-have in a standard Puppy, or are they better used as PETs?

    Administrator (BarryK)
    Wednesday, November 7, 2007, 09:06 PM
    Raffy, it doesn’t matter much if they are really small. PBcdripper is only 88KB expanded (using DU), so in a squashfs file it goes down to just about nothing. Most of the audio support utilities and libraries are there anyway.

    It’s great that PBcdripper scans for what’s there, so if extra PETS are added, like FLAC, then it gets used. I don’t have FLAC in the Dingo iso as I’m aiming for smallness, but it is good if it can just be installed then recognised by the ripper app.

    Administrator (BarryK)
    Thursday, November 8, 2007, 03:58 AM
    I mean ‘du’.

    Thursday, November 8, 2007, 08:01 PM
    I just tried both with a Steve Earle CD. I prefer Asunder. I do not rip
    CD that often. It was the easier of the two (my opinion) to use.

    Friday, November 9, 2007, 05:20 AM
    Jcoder24, I just looked at that thread on ‘gprip’. The last couple of posts are from you — it seems that there are outstanding issues that Nathan didn’t get around to fixing?

Gxset is back

November 11th, 2007

There was feedback on the forum that unable to set mouse threshold in alpha1, and I replied that Gxset as used in previous puppies is a GTK1 application and I had not yet found a GTK2 replacement. Dougal did some searching and discovered that Debian have Gxset v0.4 that does compile with GTK2, so we’re back in business.

Weird thing, I looked through the Gxset package and all of the source files were last modified in 1999 — only the ‘configure’ and associated files were modified in 2003 for GTK2. Which reminded me that Blinky and Freememapplet compiled as-is (or almost) against the GTK2 libraries. This is something unexpected, so I decided to experiment…

I tried to compile some other GTK1 apps. I failed with Xproc and uXplor, although they might have only been simple things to fix. After a tiny bit if source hacking, Usbview compiled! — works too.

The main thing in configure or Makefile is you have to replace ‘gtk-config’:
`gtk-config –cflags` with `pkg-config –cflags “gtk+-2.0″`
`gtk-config –libs` with `pkg-config –libs “gtk+-2.0″`

Usbview is a very simple display-only application. I also compiled Gtkcat — which interests me as Dingo does not yet have a CD-cataloger application. But, the text-input boxes do not accept text. Probably fixable.

One Response to “Gxset is back”

  1. BarryK Says:

    Wednesday, November 7, 2007, 05:45 AM
    Barry, with all old (=unsupported) apps, it’s a good idea to look in the Debian repos and also get their patch — they keep all the packages they host maintained, so they should compile…

Evince: a wild goose chase

November 11th, 2007

I will have Leafpad in the alpha2. This is such a nice tiny text editor, too good to leave out. I created /usr/local/bin/defaulttextviewer pointing to leafpad, as often there is a requirement for viewing text files, whereas Geany is an overkill. My ‘man’ script for example.

There are, or rather were, a couple of problems with using Leafpad.

Firstly, the ‘configure’ script is supposed to autodetect the libgnomeprintui library and if it exists then use it. Except that it fails, due to a fault in the configure script. I fixed that.

Leafpad is miles ahead of Geany with printing, and it has a nice printing dialog, also print-preview. Or rather, it is supposed to have print-preview. This is interesting, GTK 2.12.1 and gnomeprint are hard-wired to run an application called ‘evince’ to view the preview document. Evince is a Gnome multi-format document viewer, in particular PDF and PS.

I thought that I would have a go at compiling Evince, as it has configure options for disabling gnome dependency …or so it seems. It uses Poppler for viewing PDF files, which Puppy already has. Anyway, the Gnome dependency… this is where the wild goose chase started. First, it wanted ‘gnome-doc-utils’, which in turns wanted ’scrollkeeper’, which in turn wanted ‘docbook’. I waded my way through that, but then Evince wanted ‘gnome-vfs’, which in turn wanted ‘gconf’ ….aargh, I would end up with all the Gnome libraries.

So, I implemented a very simple solution. I created a little script named ‘evince’, that runs ePDFView instead. ePDFView is the lightweight PDF and PS viewer that we already have in Puppy. Works beautifully.

This is Leafpad version 0.8.12. The version used in Puppy3 is much older I think, without the sophisticated CUPS printing support.

One Response to “Evince: a wild goose chase”

  1. BarryK Says:

    Wednesday, November 7, 2007, 03:05 AM
    ref blog saga:
    Wordpress is good but uses MySQL
    If you used Flock version 1 (download and runs in Puppy 3.01) it links straight into Wordpress . . .
    I was also very happy with my Blogger account until it suddenly became unavailable - No idea why. There was a political blog dealing with Local issues, could have been that. Not sure. I imported the Tmxxine part into Wordpress.
    After much experimentation I am most happy with using a wiki as a blog, which is what I do with Tmxxine.
    Anyway that is my experience.
    Geany developer is active on John’s forum

    Wednesday, November 7, 2007, 03:23 AM
    Way to go! Evince is a nice gnome app, soon it will even have support for annotations, but still to big and too gnome dependent. Epdfview is lighter and to the point (which ist viewing pdfs). A continous view mode for epdfview would be nice, though.


    Thursday, November 8, 2007, 07:13 PM
    Is your updated Leafpad with improved printer support available to
    update Puppy 3.01 - or will it be in 3.02?

    I use Leafpad when writing HTML - yeah I like clean and simple
    manual HTML.

    I am looking forward to 4.xx as the core for my customized USB
    Stick distro.

    Thanks! doc

Dingo? Slappy?

November 11th, 2007

I chose version 4 for the RawPup, mostly because it has the latest libraries so is technically an upgrade path from Puppy3.

I intend that the Puppy3 Slackware12-compatible series will continue, so I’ll use distinguishing names. Most of the improvements to RawPup, apart from the actual packages, are readily applicable to Puppy3, so there is no trouble with releasing a 3.02 etc.
It could be called… SlackPup, Slack Puppy, Slappy?

I’m thinking that the new series with latest of everything and GTK2 only, needs a better name. RawPup doesn’t convey much. I’m thinking that I might announce this new pup when it gets released, in the local monthly “Perenjori Bush Telegraph” newsletter — they charge only 15 dollars for a full page advert, which gives you an idea of the extent of its circulation! Anyway, if I’m going to promote it locally, I thought an Australian name would be appropriate, and ‘Me’ suggested ‘Dingo’ (seconded by ‘zygo’).
Yes, ‘Puppy Dingo version 4.00′ has a nice ring to it. or, the other way round, ‘Dingo Puppy’, which we will come to know as just ‘Dingo’.

A Dingo incidentally, is a native wild Australian dog, apparently brought to this land 40,000 years ago by the Aboriginals.

Sigh, this new blog script has a bug. The comments section on the right side isn’t updating. This new blog is still on trial, and I might move on. If I do so, I’ll copy all the posts and comments.

One Response to “Dingo? Slappy?”

  1. BarryK Says:

    Tuesday, November 6, 2007, 06:24 AM
    ” I’m thinking that I might announce this new pup when it gets released, in the local monthly “Perenjori Bush Telegraph” newsletter — they charge only 15 dollars for a full page advert, which gives you an idea of the extent of its circulation! ”

    I like it a lot…….please send me the bill if you do that….
    and I’ll happily Paypal you.

    But no pics please of you wrestling a Dingo in the local pub…
    a la Crocodile Dundee.
    We don’t need too much excitement for our new users to be from MS.

    Showmanship and Vista…..or rather slowmanship with Vista….( that is not a typo….that is a P )…….has taught us that.


    Tuesday, November 6, 2007, 08:53 AM

    Naming RawPup as Dingo has a nice ring to it. Besides, people around the world should know Linux is also being made Down Under.


    Tuesday, November 6, 2007, 11:56 AM
    We vote SlackPup and DingoPup for versions 3 & 4.

    Tuesday, November 6, 2007, 01:29 PM
    Names sound all good (Dingo and Slappy). Apropo blog… have you tried blosxom ( http://www.blosxom.com/ )?

    Will likely test the beta.


    ps. I missed to post about gview and Fotox (because of the ever changing blog ;)… Ever heard of gpicview ( http://lxde.sourceforge.net/gpicview/ )? Nice, fast and little image viewer with UI resembling the WinXP image viewer.

    Tuesday, November 6, 2007, 03:53 PM
    How about Kelpie Puppy for v.301 ?

    The Kelpie is a Australian dog that according to Wikipedia “has proven very successful at herding and droving with little or no command guidance”.

    Tuesday, November 6, 2007, 04:21 PM
    Have you considered Wordpress for your Blog? Take a peep at http://www.wordpress.org. If you download XAMPP you can host it yourself, or your ISP could (they’re certain to have mysql, per & php. Or, Wordpress themselves can host it at http://www.wordpress.com.

    XAMPP & Wordpress both work fine under Puppy 3.01.

    I like Dingo; especially as they aren’t a bit tame!

    Tuesday, November 6, 2007, 05:58 PM
    okay the firewall is not working - it is complaining about components not having been compiled

(c) Copyright Barry Kauler 2008, all rights reserved. http://puppylinux.com/