logoPuppy developer news:

from 2.17 to 2.17.1

left-arrow Older news

arrow-rightLater news

Puppy 2.17.1 released 

This is a bugfix and minor-tweaks upgrade of Puppy 2.17 -- read earlier blog posts for details.

The primary download site is:
http://distro.ibiblio.org/pub/linux/dis ... uppylinux/
Some people report difficulties downloading from ibiblio, so one of the ibiblio mirrors may suit you better:

Good news for people on dialup. This Puppy is 94MB, but I have split it into 5 parts that you can download individually. There is a script for putting them together (and it does an md5sum check of each part) and a README with further details. See here (also the mirrors):

If you have 2.17, there's no need to upgrade, unless there is some issue that I have mentioned in the blog posts since the release of 2.17 that affects you. The non-recognition of the USB keyboard is one example, but that is a one-off problem that only occurs on the first boot so even if you had that bug it might not concern you anymore. The Universal Installer may behave better for some people -- but then it can be downloaded from the forum, you don't have to get an entire new 2.17.1 iso.

The Network Wizard is unchanged from 2.17, but the 'dhcpcd' package and a few wireless drivers have been upgraded, which fixes some problems. If you still experience problems with the Network Wizard, there is a new experimental version available via the forum:

I will upload the 2.17.1 Unleashed core package, plus some updated PET packages, in a day or two.

Release notes for Puppy 2.17:

Upgrading from Puppy 2.17:
Don't forget the usual thing -- if you have used 2.17, Puppy does not recognise the '.1' appendage so doesn't know this is a new version.
So, you have to bootup in ram and delete the pup_217.sfs and zdrv_217.sfs off the hard drive. Then boot 2.17.1 cd with 'pfix=clean' (or set /etc/puppyversion back to '216' before rebooting).

Works fine using frugal install and pup_save.2fs file created with Puppy 2.17 and keeping /etc/puppyversion unchanged.

BitTorrent is a nice alternative for those with limited internet speed as well because of the resume and integrity check feature of most clients. We are hosting a Torrent for the latest release here: http://www.tuxdistro.com/browse.php?cat=284

The remaster script is working OK. I was able to put together an ISO based on this Puppy in a few hours without much effort :)

Blank screen on an Acer Aspire 5630 Laptop with Nvidia GeForce Go 7300 and 1.66GHz Intel Core 2 Duo CPU. I had the same problem with the previous buggy version of Puppy 2.17. It seems that it is not fixed in the new version 2.17.1. I assume the problem occurs when I am about to choose the keyboard language.

Somebody has reported this already, I can confirm.
Upgrading from puppy 2.16. I got two icons "drives" on the desktop.

I've just discovered that my (not often used) wireless card has no kernel module in 2.17.1. It didn't in 2.17 either but did in 2.16.1 and 2.16:- DESCRIPTION: D-Link System Inc DWL-510 2.4GHz Wireless PCI Adapter VENDOR: 1186 DEVICE: 3300 KERNEL MODULE: r8180

Also, it makes a lot of sense to have only a nolzma iso uploaded. But it must need more ram to run in (from live cd without hard disk). How would I convert it to a lzma iso?



The rtl8180 wifi driver isn't supported with the linux kernel yet. There is an active project that is supporting the card, with an aim to get it in to a future kernel.

alsaconf-parameters (ES1869 on board) are not kept after reboot

work fine in 2.16 en earlier.

Puppy work great on my PII 333MHz with 128MB ram!



How is rtl8180 related to r8180?

The network wizard called a rtl8180 a r8180.

There seems to be a bug with keymaps, when using xvesa. That is what I've just discovered.

There seems to be always the US keymap instead of the DE, which cannot be changed. When you switch to Xorg, everything is ok.
This is also reproducable with puppy 2.16

Barry, could you please fix this in 2.17.2


WARREN S CARL (windowsystemcomputers<at>yahoo.com) 
$150 7" Laptop PC
Mini PC with 7" LCD Display
TFT LCD 16x9 Res. 800x480
Fanless 150-300Mhz CPU,
Audio, 64mb ram, 1GB Flash
Lan, 6 USB. Option: wireless
Linux, browser, email, etc.

UPDATE: util-linux fdisk restored 

I have decided to stay with the 'fdisk' utility from the util-linux package. The boot script rc.sysinit runs 'fdisk -l', and the Gnu version is extremely slow -- five seconds to return results on my laptop, compared to virtually instantaneous with the older fdisk. I don't want to slow down the bootup that much.

I will however keep the new Gnu cfdisk.

fdisk, cfdisk utilities updated 

Puppy has used the 'fdisk' and 'cfdisk' utilities from the util-linux v2.12q package, however we have just discovered that cfdisk is broken. I also have some suspicions about fdisk.

I found that there is another fdisk/cfdisk package at gnu.org, not part of util-linux, that is actively maintained, and I have upgraded Puppy to use it. The Gnu fdisk has more features than the util-linux fdisk, but I did a quick check of 'fdisk -l' and the output format looks the same, so it should not break any scripts (actually, 'fdisk -l' gave a lot of output on stderr, whereas the util-linux version gave no output on stderr, so that is a difference).

The home page for this package is: http://www.gnu.org/software/fdisk/

modprobe bugfix 

Thanks to TonyVegas who discovered a problem with module loading on the very first boot, in which one or more modules failed to load. This bug only occurred before a pup_save file existed. He worked out a fix for the 'zdrvfind' script. Based on his work I have applied a fix to 'zdrvfind' and 'modprobe' scripts.


A couple of apps reverted 

For 2.17.1 I have changed a couple of applications.

Have dropped Grisbi and replaced with HomeBank v3.2.1. Geoff has been testing Grisbi, and it has too many nasty surprises. HomeBank is less-featured, but note that v3.2.1 is not the latest -- the latest stable is v3.3 but it requires GTK 2.10.x whereas Puppy currently has GTK 2.8.x. I plan to upgrade GTK for pup 2.18.

Xfdiff, the frontend for 'diff' and 'patch', is buggy. I have informed the author, but for now have dropped back to TkDiff.

Tkdiff is superior to Xfdiff, anyway. It not only shows the lines that are different, but the difference within those lines.

the upgrade to gtk 2.10 by v2.18 thats great that means that compiz-fusion will be compatable with puppy now thanks

"...but for now have dropped back to TkDiff."

Yay! Tkdiff is a great differences utility, intuitive to use.

CUPS print-to-PDF updated 

jcoder24 has updated this package to version 0.3:

I have put this into Puppy, so it will be in 2.17.1-final.

Puppy 2.17.1rc uploaded 

See further details in my previous Blog post.

Get it from here:
http://distro.ibiblio.org/pub/linux/dis ... inux/test/
Here's the md5sum:
18b05ae816d4bbd47539e33a80923645 puppy-2.17.1rc-nolzma-seamonkey-fulldrivers.iso

Well, that was an interesting experience... up until now, I had never used my satellite connection (I got it in February this year) to upload a large Puppy file. Always I had gone to my relative's place who has ADSL2. My satelite download speed is 512Kbps, upload only 64K, however what I discovered is that upload is only about half that, effectively about half of dialup speed! You can imagine, thousands of packets hitting the satellite simultaneously from a thousand different places on the earth -- how does the satellite sort out that mess? What I observed is that it goes through a cycle of negotiation then a burst of data transfer. So, Gftp would show 'transfer stalled' while blinky showed furious negotiation going on with the satellite, then suddenly Gftp would do a data transfer burst, then the cycle would repeat.

Consequently, to upload 93MB.... yikes! I broke it up into five parts, and watched the first part (20MB) upload in about 3 hours, then I went to bed. This morning it was all uploaded, and I logged in to ibiblio and put the files together. md5sum checks out okay.

This is good for people on dialup! You can download Puppy in small pieces. There is a file 'parts.md5' that has the md5sums of each file. To put it all together:
# cp puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso.part1 puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso
# cat puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso.part2 >> puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso
# cat puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso.part3 >> puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso
# cat puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso.part4 >> puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso
# cat puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso.part5 >> puppy-2.17rc-nolzma-seamonkey-fulldrivers.iso
# sync

If this split-files thing is popular, in future I'll create a little script to do the reconstruction.

Don't forget the usual thing -- if you have used 2.17, Puppy does not recognise the '.1' appendage so doesn't know this is a new version.
So, you have to bootup in ram and delete the pup_217.sfs and zdrv_217.sfs off the hard drive. Then boot 2.17.1 cd with 'pfix=clean' (or set /etc/puppyversion back to '216' before rebooting).

...this is all a bit tedious, I'm thinking I should work out how these sub-versions can be treated as new versions. Maybe use four digits for the version number: 2171.

Barry, I personally would really appreciate the split files thing as a future practice... it would be great!

ftp sits on top of tcp. The delay on the satellite link (around 700 ms round trip) means that the whole transfer process slows down even with a high bandwidth potentially available.

Forum thread for feedback:

Ron (newsgrou<at>gmail.com) 

Read up on delay-bandwidth-product. With high available bandwidth and high latency networks, opening multiple connections gives you a bigger piece of the pipes bandwidth. When I had one-way DirecPC (modem outbound satellite inbound) , I could sometimes get download rates of 2.5-3.0Mbps and I way only paying for 256Mbps. I used Download Accelerator and configured it to open 5-7 connections. With the standard single ftp connection the download speeds were quite slow. Not sure if a FTP client exist to perform the same function in the upload direction but you could break the large ISO file into several pieces and perform a upload test. Using your sat uplink, you might only be allowed one FTP connection? With the fairly low bandwidth of your uplink the results may not be as dramatic, but you might get closer to your 64kbps.

When I had my one-way satellite access I got some good information from the following web site.

Good Luck

PS.. If you test your download/upload rates with multiple connections, please report back with your results.

The dog in 5 parts is very welcome. I have a very variable download speed at work. Sometimes an iso takes minutes sometimes hours. The trouble is that the dl is copied to my home drive only if it completes on the gateway. Sometimes the day ends and I have to abandon the dl. So I loose everything.

Ted Dog 
cat puppy-2.17rc* >pup217rc.iso

Ron (newsgrou<at>gmail.com) 
Paying for 256Mbps in my previous post should read 256Kbps.


Universal Installer: making a partition bootable 

Someone reported on the forum that when installing Puppy to a USB Flash drive, at one point there is the message "Making partition bootable...", which that person pointed out is a lie. Yes, it seems that when Syslinux or Extlinux installs in the partition to make it "bootable", that is different from the 'boot flag' of the partition itself -- which must also be set. Well, I think this is the situation anyway.

I have been messing around doing installs to USB Flash, and encountered a few interesting things. I found that 'fdisk' gets confused -- I had couple of situations where fdisk reported spurious partitions, where GParted and disktype reported correctly only one partition. So, I don't trust fdisk anymore, and have reduced its usage in 'puppyinstaller' to only one place (which will be considered for removal sometime).

Regarding making a partition bootable, I added an extra check, and if the 'boot' flag is not set, GParted is started, with instructions given how to set the boot flag. Note, previously I might have done that directly from the commandline using fdisk, except that I don't trust fdsk anymore. But, it's very easy for the user to do the few mouse clicks required to set the 'boot' flag in GParted.

Not sure if cfdisk has the command line parameters required - but just in case you forgot that is in Puppy


cfdisk --h
cfdisk: invalid option -- -

Print version:
cfdisk -v
Print partition table:
cfdisk -P {r|s|t} [options] device
Interactive use:
cfdisk [options] device

-a: Use arrow instead of highlighting;
-z: Start with a zero partition table, instead of reading the pt from disk;
-c C -h H -s S: Override the kernel's idea of the number of cylinders,
the number of heads and the number of sectors/track

Mr Doolie 
Thanks, Barry. That person was me. The first time I ever used the installer I ran through it like 5 or 6 times. "Making partition bootable". Nope. It won't boot. Try again...and so on. Discovered the boot option of fdisk by accident and was able to fix the problem from then on but it sure was confusing to a Linux noobee.

Thanks from me too.
It was never possible for me to boot from USB Flash. Reading through the Forum of all sorts of problems and strange BIOS's, there were even some there had their USB Flash wiped out - I gave up. Looking forward to try again.

Barry, fdisk can't handle superfloppies. I've noticed in many HW-test results that it jsut gives you that funny list of "partitions" and the message about "this doesn't look like a partition table".

I use disktype -- it's also useful for getting the fs type (fdisk shows both ext2 and ext3 as "Linux".

Hi Barry et al,

My way of doing usb bootable install is as follows,

I use Gparted only and format the usb flash to ext3 (it's what i use).

Then when complete, go to manage flags, and select boot and apply.

Then use the Universal Installer to load Puppy and choose to leave mbr alone...do nothing.... and no worries.

Works everytime for me....


Eric/caneri (ericmulcaster<at>gmail.com) 
To add to the above post..

Flash drives with U3 on them will not work on booting Puppy...

U3 must be removed before Puppy will boot...because U3 is a boot partition in itself.


Dougal, I had a situation in which the Flash drive had one partition, sda1, which Gparted, probepart and disktype showed correctly, but fdisk showed lots of spurious partitions. That Flash drive had been converted backwards and forwards between superfloppy and various permutations of partitions on it, so there must have been some residual from its earlier superfloppy status that upset fdisk. I recall, I cleared the boot flag, from Gparted then set it again, then tried fdisk again and suddenly it showed the partition correctly.

Eric, what is U3? I haven't encountered that before.
How would the Uni Installer recognise it?

Annologic (annodevries&lt;at&gt;gmail.com) 
Hi Barry (and others)

The U3 feature is a "booting" environment, also for usb sticks. Sandisk calls it "smartdrive". read all about it here:


I think we could do without, probably do a reformat first before any install on a stick with this software on it( don't know, probably all newer types?)

eric/caneri (ericmulcaster<at>gmail.com) 
Hi Barry,


U3 needs a special removal tool to get rid of U3 programs and partition. I have it but need to send it by email or it's on the url below

http://www.sandisk.com/Retail/Default.aspx?CatID=1415 for Sandisk flash

http://www.u3.com/uninstall/ this is for all other usb flash drives with U3

I have tried to load Puppy on U3 drives..they will accept the pup but will not boot.

Hope that helps...I tried to explain U3 on the chat but not many have it...this will be around more and more so this will be needed to use a usb flash.


I have played with a few Puppies since 2.12. When the test kernel "Cutting-Edge-Puppy" was posted in the Cutting Edge section of the forum I attempted an install the a clean ( wiped to zeroes ) hard drive.

I had no problem concerning the partitions ( swap & EXT3 ) or making it bootable, I did spend 2 days attempting to get the devx_217.sfs to work when it had been working fine on my USB stick.

It seems the SFS files are not 'unioned' ( for lack of a better understanding ) when a hard disk is dedicated to Puppy. This differs from what appears to be a 'frugal' install to a hard drive ( which I didn't try or learn about until stumbling across a forum post ). I could rename the devx_217.sfs to usr_devx.sfs and they would be processed at startup but wound up in /usr/usr/bin instead of /usr/bin ( was wanting GCC really badly ).

The above mentioned forum post ( frugal vs full hard drive ) included instructions for mounting the devx_217.sfs file and issuing a 'cp -ax xxxxx xxxxx' to install the development files in the correct directories.

After all that explaination, I guess my question or suggestion is:

Why doesn't the hard drive install act just like the USB install? I mean if I want to add one of the SFS bundles, why can't I just put it in the / directory and Puppy process it like it does in a USB install?


ps - the SMP kernel seems to work fine for my purposes. I don't know all the constraints but wouldn't a SMP kernel work well on a 1 CPU box as well as a 2 CPU box? That way you only have to package 1 kernel ( To my suprise a Vector 5.8 install on a 1 CPU box booted a SMP kernel on a 2 CPU box after moving the hard drive to the new box ).

Network Wizard improved 

Tempestuous and Dougal have been working on this. I have put their latest version, dated July 28th, into Puppy. The 'net_setup' PET package is updated to version '2.17.1' (to match the Puppy version it will be released with).

If you would like to test it, see this forum thread:

Puppy user 
Using 2.17 on my Acer Aspire 4310 laptop, the wireless adapter was detected automatically.

There's a hiccup here, see this thread:

Currently, this is all I have on my list to fix before uploading 2.17.1. Anything else can wait.

Puppy 2.17.1 coming on the weekend 

2.17.1 is supposed to be bugfixes and some tweaks applied to 2.17, but the longer I delay the more things get changed and the more chance of new bugs appearing. I can't release it right now, as although I have a satellite Internet connection it has very slow upload speed (only 64K). At the end of the week I'll be at a relative's place in Perth, with ADSL2 speed, and will upload 2.17.1-rc then. As I'll only be there a couple of days, I will have to upload the 2.17.1-final a couple of days after the 'rc'.

If you have 2.17, there's no need to upgrade, unless there is some issue that I have mentioned in the blog posts since the release of 2.17 that affects you. The non-recognition of the USB keyboard is one example, but that is a one-off problem that only occurs on the first boot so even if you had that bug it might not concern you anymore. The Universal Installer may behave better for some people -- but then it can be downloaded from the forum, you don't have to get an entire new 2.17.1 iso. Ditto for a few other things, like the Network Wizard and some updated wireless modules.

Please can you look at remastering

Dougal's took a long time and produced no iso (as is my preference). I stopped the other remaster script after 10 minutes of decompressing the sfs.

Also had problems using both remaster scripts (Dougals and standard)

I think the problem is not the script but the mksqashfs in puppy 2.17. But that's what I've already mentioned in the thread zygo refered to above.

John Biles (jgvg9<at>bigpond.com) 
Hello Barry, what are you thoughts on this bug in Puppy 2.17 preventing the remastering of new iso's with extra packages added.

I just surprised with this being a major bug to alot of the forum user's that no comments from yourself have been posted.

Even a "I'm Looking into it" to let us know your working on it would be good.

There's a bug in the new mksquashfs with LZMA patch -- it is supposed to accept a commandline option '-nolzma' to build a file without LZMA, but it will not accept that option. Creating a file with LZMA takes a long time.

Okay, I'll use the older mksquashfs from Puppy 2.16, and rename the new one as mksquashfs-lzma.

You can test if the old one fixes the problem, just grab it from 2.16 -- the older PET package should be on ibiblio.

How MS stays top dog 
My first thought when I read this is "that person is brave, accusing MS of breaking the law". Below are extracts from http://www.desktoplinux.com/news/NS9526259887.html:
<quotes deleted, see link>

" The agency had petitioned U.S. District Court to find Microsoft (MSFT)in contempt for not providing an adequate way of separating Internet Explorer from Windows 95. It had said previously that bundling the browser with the operating system was a violation of the 1995 Consent Decree signed by Microsoft and the U.S. government. '

Shortly after MS released Windows 98, another Aussie bright spark
Shane Brooks, made a mockery of Mcrosofts' BS with 98Lite.

Many whinged about 98 and 98SE - but if you used Shane Brooks program
and knew what to do ...and what programs to avoid....the Blue Screen of Death rarely occurred. And I used to run over 200 games.
Still had Dos and the wonderful QuikMenu III.


I had low spec machines Pentium 11 100 computers with 8 megs
running easily.


98lite is a utility for Windows 98 and Windows Me written by Shane Brooks. It removes Internet Explorer components from Windows along with several operating system components that require IE to be present. 98lite was one of the first programs to provide a method for removing Internet Explorer. It was written in response to Microsoft's claim, in antitrust proceedings, that IE is integrated into Windows and cannot be removed without breaking other features.

" He discovered a way of removing IE manually and presented it to the United States Department of Justice as evidence that Microsoft was monopolizing the browser market."



There you are Barry....he might give you some clues as to fossicking
for gemstones. Anyway he was a genius...as are you.



" When Brooks first told people what he had done, they didn't believe him. So he set up a World Wide Web site (www.98lite.net) to explain it. Then he put his program on the site to allow others to download it. More than 60,000 people have done so, according to a counter on the site.

His program has won largely favorable notices in the computing press, both as software and for what it says about the government's case against the software maker. The Justice Department is contending that Microsoft included its Internet browser with Windows 98 to drive competing browsers out of the market.

"So easy that even a federal judge could do it," Brian Livingston wrote in InfoWorld. After 200 readers related their experiences, Livingston concluded that "Overwhelmingly, readers who ran Brooks' method to remove IE from Windows 98 reported that their systems ran faster and with fewer crashes."

"In an interview, Livingston, the author of "Windows 98 Secrets," said that "Brooks is an embarrassment for Microsoft."

John Breeden of Government Computing News, a trade publication that is owned by The Washington Post Co., gave 98lite a generally favorable review in The Post, calling it "PC liposuction."

"During the trial . . . Microsoft kept saying, 'There's no way this can be done,' " Breeden said. "But here's an example of what is basically a kid, a programming amateur, who has done it in his free time."

Brooks is not quite a kid -- he's 32, married and with a young daughter. But he definitely is an amateur programmer, in the sense that his true profession is ecologist. He currently is a post-doctoral student specializing in river restoration.


Sour grapes.

Microsoft wins because it delivers what most customers want. Not what I want - I'm so glad Puppy is here and has delivered me from what felt like PC-based torture. But what most customers want.

Microsoft can also do what it wants with its own property, including selling "below cost" and all the rest - if property rights mean anything (which I think they should).

The anti-trust action was a blow against a company that worked entirely in the market. Now Microsoft has lots of high-paid lobbyists in Washington DC, and probably owns a few Senators and Representatives too. This is progress?

Please note that almost all so-called anti-trust and regulatory actions in the US have the effect of cementing the positions of established businesses in their markets, and thwarting competition from smaller competitors. In other words they do the opposite of what they are claimed to do. Welcome to the world of corporatist government.

Go to lewrockwell.com and search for "Microsoft" to get a non-PC non-socialist view about this.

The way to beat Microsoft (whose products I hate, and whose computing environment I hate) is not through government, but in the market. Just as Puppy and PCLinuxOS is starting to do right now - there is a huge potential market for Windows refugees out there. Give it time...

In one thing Con Kolivas was right.
In One thing Con Kolivas was right

"However, the desktop PC is crap. It's rubbish. The experience is so bloated and slowed down in all the things that matter to us. We all own computers today that were considered supercomputers 10 years ago. 10 years ago we owned supercomputers of 20 years ago.. and so on. So why on earth is everything so slow? If they're exponentially faster why does it take longer than ever for our computers to start, for the applications to start and so on? Sure, when they get down to the pure number crunching they're amazing (just encode a video and be amazed). But in everything else they must be unbelievably slower than ever.

Computers of today may be 1,000 times faster than they were a decade ago, yet the things that matter are slower.

The standard argument people give me in response is 'but they do such more these days it isn't a fair comparison'. Well, they're 10 times slower despite being 1000 times faster, so they must be doing 10,000 times as many things. Clearly the 10,000 times more things they're doing are all in the wrong place."

This is the reason I have moved from OpenSuSE to PuppyLinux.

Keep moving forward.

Thank you and all the best


I was just about to mention that Con Kolivas interview!

Barry, it's definitely worth a read.

First let me give my background: Degree in the hard sciences, coursework that included Donald E. Knuth's "Art of Computer Programming" (I wonder if it still have a picture of Tom Mix on the cover!), etc. Have done a signficant amount of application programming, however, I would not say I'm a professional programmer. With those brief words serving as an introduction, I say these words that I hope will offend the Linux crowd (yes I hope they offend) because maybe you'll wakeup and growup. Windows is better than Linux (Puppy, Ubuntu, Knoppix, DSL, DSL-N, etc...) --- I've tried them all --- the linked article includes one the many good reasons why:

...while OpenOffice.org can do most of what Microsoft Office can do, it can't run Office's VBAs (Visual Basic Applications) and macros, and since businesses depend on these, blah, de blah, blah, blah....
There's only one problem with this argument. It's not true. Thanks to Mono, you can run VBAs in Linux. In addition, Novell and Microsoft are specifically working on making it easy for users to migrate Excel VBA macros to OpenOffice's Calc. ..

I make very extensive use of Microsoft ActiveX capabilities to automate other advanced numerical analysis programs so that they work *beautifully* together saving vast amouts of time and money. It is irrational to think Linux is remotely close to Windows in these areas that the author of the linked article so idotically dismisses with "... blah, de blah, blah, blah".

By the way, I've tried to convert some of my processes to OpenOffice's structure --- please, do try to do this for yourself for a non-trival business application and see how poor OpenOffice is compared to Excel/VBA. Just see how well OpenOffice work with all othe ActiveX applications. I have tried this --- about a year ago --- unless things have improved dramatically, it was *terrible* then. And bringing up the rear and trying to be as good as (or even better than) Excel/VBA several years *LATER* violates the maximum of the famous and brillant Confederate General Nathan Bedford Forest said was key to his many victories: "Be First with the Most" --- and Microsoft also applied this maxium brillantly in its David vs. Goliath battle with IBM at the dawn of the personal computer age.

I enjoy playing with Puppy (as well as a few other interesting Linux distribution like NimbleX and even DSL) and Linux has made great strides in being competitive on your basic home use of surfing the web, writting letters, simple spreadsheets, backing up data, manipulating files --- but Linux is simply no match in real world business applications the way Microsoft is. Bill Gates/Paul Allen/Steve Balmer/etc. are worth billions and they deserve it, they are very, very, smart people, worked very hard, and yes they did, and still do imploy some very agressive cut throat methods that I don't like --- welcome to the real world boys and girls, wakeup and growup, you have got to "Be First with the Most" not some time years down the road, time waits for no man (or woman).

Hmmm, I've strayed off the track, with a "political" post. I should stick with what this blog is supposed to be about.

Yes, I agree that there is a lot of good stuff in Windows OS and apps. But, this is a subject that we could argue about for a very long time, with endless points to be made one way or the other. I guess what it comes down to for me is I don't like MS's near-absolute monopoly -- it's fundamentally unhealthy. I also got tired of giving my money to MS. I've been in the computing field for a long time, and gradually developed a digust for MS's tactics -- it's really down at the lowest level -- but of course you can take the point of view that they are pretty clever people to get away with it and choose to amire them for that -- but it's a perverse kind of admiration.

So for me it's an ideological issue, for which I'm prepared to put up with some shortcomings, with for example, OpenOffice. But, there's a lot of us, and we keep plugging away, steadily improving everything. Of course, I could just go and by a Mac (I have owned a few of them by the way).

Amyway, let's consider this thread closed.

Anon, which percentage of Windows users actually use macros? 1%? 2%?
I don't know anyone who does.
So that might justify 1% of computer users using Windows.

Sorry to come late to the party. After reading Con's article I would suggest that Barry apply Con's CK1 patches to Puppy and make him even FASTER. Puppy will always be a desktop OS and it would be an encouragement to Con that his work is appreciated and adopted. Who knows - he might get the fun back into his coding again!

True flushing of RAM to pup_save 

This is discussed a few posts down this blog. Andrei has developed /usr/sbin/snapmergepuppy-flushing to handle true flushing when the commandline has a file with list of files to be flushed.
He has also created /usr/sbin/snapping, which is a "USB Flushing Wizard", that can configure how flushing works.

I have integrated flushing into PETget, the package manager. When packages are installed using PETget, a list is created of all installed files, and /usr/sbin/save2flash is executed (the same executable that is run if you click the desktop "save" button) -- this sends a request to /usr/sbin/savepuppyd to perform a flush.

savepuppyd is a daemon. When it sees the request, if Puppy is running in the correct PUPMODE and if the list of files to flush exists, then /usr/sbin/snapmergepuppy-flushing is executed with the list-of-files passed to it.

Note, the reason for the request mechanism via savepuppyd is so that snapmergepuppy* does not get called while another instance is running. All executions of snapmergepuppy* are via savepuppyd, if click the manual desktop save button or install a package with petget.
The exception is at shutdown, when /etc/rc.d/rc.shutdown will call snapmergepuppy* directly.

I have set this up, but not yet tested it...

Wow Barry and Andrei, nice refinement of the save mechanism.

There is another case, that I ran into. When getting large amounts of mail, I have seen my memory go to zero. I wonder if the Seamonkey mail folders can be copied down as you go? Each mail should copy into a different location so we shouldn't be wearing out the flash.

I bet this one is harder to implement, and we'd have to worry about other mail clients too.

Of course the question is, why is a mail download consuming 80MB of memory? I don't know, doesn't seem right.

Please update e2fsck (e2fsprogs) to 1.40(.2).

Ramengo, what's so urgent about updating e2fsprogs? I've actually already done that for myself, but am just curious what's the big deal about it...

I'm sorry to bore you but it's almost two months that I'm trying to change my old PC partitions. My situation is similar on described on http://bugzilla.gnome.org/show_bug.cgi?id=446778, but I can reassume it here briefly:
I use PuppyLinux to manage GParted to reduce the size of my first partition and reallocate the space to the second one. GParted live cd doesn't even start on my very old PC, that's why I'm using this powerful (!) PL.
The problem with GParted shipped with PuppyLinux is that it uses old e2fsprogs routines (ie e2fsck) and so it can't grow the second partition correctly.
I hope that upgrading to latest e2fsck will resolve my problem.

You should be able to compile the latest e2fsprogs and use it in Puppy.

Regarding true flushing of RAM, probably not in 2.17.1. We still have some problems to sort out.

ramengo (ramengo<at>hotmail.com) 
Sorry, but I've never compile anything on Linux...
Do you have any hint to suggest, please ?

Ramengo, I'll post what I compiled on the forum.

Thank you very much !
Please post here the link to the thread too ;-)

Here you go:


PETget bug fixed? 

Hock-Teck Chia made an interesting discocery:
I found that by changing the environment
setting using "export LANG="en_US"" and starting
pupget from the console, I can get the pupget to work
without the window disappearing. (My original setting
was "LANG=en_SG".)

So, if you have a problem with the text not appearing in the 2-pane window in the PETget package manager, what locale do you use? If something other than en_US or en_AU, test PETget by running it from a terminal, firstly normally to see the problem, then export LANG=en_US:
# petget
# export LANG=en_US
# petget

...let us know if it works! Note, if it does, it is a bug in Xdialog, not PETget.

I had a case of my window disappearing recently. After looking around I found that what had happened was the lower window bar had somehow been pulled all the way to the top, squashing the window in a thin line. I don't know if it is related to this, I don't think it was in the petget mgr. It still showed up in the taskbar.

I think this is a different issue.

Problem 1):
no window (but a Xdialog -box displaying the "usage"

Problem 2):
An empty pane on the right.

1.) This can happen like this:
you have a different locale.
This will throw a warning to stderr: "warning: locale not supported by libc".
As Xdialog uses stderr to pass return-values, the returnvalue now is messed up with the warning.
This also can happen with broken Gtk-themes (referring non-existent icons), what also throws warnings on stderr.

Now, if your code expects a return-value like "1", and then runs something
Xdialog -$returnvalue
then Xdialog does not run:
Xdialog -1
Xdialog warning: locale...

This is an illegal parameter, hence the window does not appear. Instead the "usage" is displayed.

Solution: export LANG=C in the beginning of the script.

2.) It was reported, that when the second pane on the right is empty, also /root/.packages/.alienpackages.txt was empty.
So it seems, this is a different error.
Somehow petget saves an empty alienpackages.txt sometimes.
It had to be found out, why.


[blockquote]Solution: export LANG=C in the beginning of the script.[/blockquote]
Okay, done.

this thread reports a broken right pane, too:


Here's a few links..

Calls to 'dillo' replaced with 'defaulthtmlviewer' 

Puppy 2.17 has 'gtkmoz' as the internal HTML viewer, replacing 'dillo', and this has exposed a problem with two packages that call dillo explicitly to display their help pages.

Forum member 'mmm' reported that Ian's little Xwget has this problem. I fixed it and changed the version number from 0.5 to 0.5.1.

Nathan has fixed the same problem in 'freememapplet' (that you see in the taskbar), and has bumped the version from 1.2.1 to 1.2.2.

It doesn't matter if we go back to using Dillo. Whatever the internal HTML viewer is, 'defaulthtmlviewer' is a script that provides the indirection to launch the correct viewer.

For the masses, these "default" files are located in /usr/local/bin/ (at least in Puppy v2.11) and are easily changed with a basic text editor.

HardInfo: system information 

I have posted hardinfo- PET package here:

One problem, when I click on Devices --> Storage, get a segmentation fault. I would like to know if that is just my hardware or if others get that crash.

Other than that, it looks good.

When I click storage, hardinfo simply closes. Benchmarks look pretty good, as far as I know ;-)


Why not use System Stats, as in DSL? Still my favourite.

I compiled it a few days ago and it worked fine for me.
Going to devices gives me a list of my drives and info about them.

Could it be something to do with the new kernel? I'm still usign the old one.

I'm using 2.17 full hd install. Hardinfo seem to work very well, though it did seem to tax the system resources when I generated a report. I thought the machine had locked up for a few seconds. the report was generated OK though. I have a 3 ghz P4 and 2 gig of ram.

Oh by the way clicking on storage also works :)

Grndoor (grndoor<at>blueyonder.co.uk) 
I am using 2.16 and 2.17 full hd installs.

I have tested "hardinfo" from the two "pet" packages and also compiled it locally from the latest svn version.

"Devices --> Storage" works fine on three different PCs, including a Compaq Proliant 1600 which is an all SCSI hard disk machine.

However, whenever any USB storage devices are plugged in, running "Devices --> Storage" produces a segmentation fault.

The USB devices included memory sticks, a card reader and an external cdrom/dvd box.

Details on how to run Puppy full hd installs on the scsi hd machine, using a boot initramfs file, will be posted later.

Eric/ caneri 
Hi Barry et al,

I crash on Devices.....Storage

HP nx9420 lappy....Intel core2 duo,1gb ram

running 2.17 with ezpup

looks good for me and will make sys info easier for everyone I think.


Yes, crashes on me too. My SATA drive shows up as SCSI. I don't have any IDE drives. Barry are you usimg a SATA drive?

Installed hardinfo.pet. No problems including NO segmentation fault message. Interestingly I am now using Australian English. Always had a secret desire to be an Aussie. Also liked the "Benchmarks". All "Benchmarks" worked EXCEPT for "CPU ZLib".

Under "Display" my video mfgr was not recognized. Under "Filesystems" all my partitions were recognized. Under "Storage" all my HDs were recognized as well as two USB devices (USB cardreader and USB flash drive). "Input devices" correctly identified my keyboard and Logitech mouse. Neither is wireless.

What is "Boots" info for??

I am using Puppy version 2.16 as a live DVD with pupsave file on HD. System info is pentium 4 3.0ghz with 1 gig memory and two harddrives.

Universal Installer feedback needed 

This is a request for feedback only if the Universal Installer does not recognise all the partitions on your hard drive. This has been reported by a few people while testing alphas and betas for v2.17 and also the final release. However, the feedback has been very inconsistent -- one person reported that the same problem existed for v2.16, while a couple of others reported that the Universal Installer in v2.16 recognised all partitions but v2.17 did not. One person reported that after he used GParted to re-format the partitions, the problem went away. I think that the problems were reported with IDE '/dev/hd*' drives, with the usual ext2 and ext3 filesystems.

I have been testing the v2.17 Universal Installer on any PC that I can find, not doing an actual install, but just going far enough to see what partitions it recognises. Guess what, it has been correct every single time. So, how can I fix a problem that does not exist?

The fact that someone used GParted and the problem went away is a clue. My first thought is that there could be a mismatch between the partition ID number and the actual filesystem. So, for people who have this problem, could you please do some tests to look for an inconsistency:

# cat /proc/partitions

# probepart

# fdisk -l

Run 'puppyinstaller' and report what partitions it recognises.

Here is a forum thread to provide feedback:

Warren S Carl (windowsystemcomputers<at>yahoo.com) 
The Problem is not that the Uni-installer can not see bare HD's. The problem is that puppy 2.16,2.16.1 and 2.17 cd boot can not see HD drive with puppy installed. And yes reparting is a work around, but not an upgrade! I had a laptop with 2.15 (gparted HD) and 2.16 uni-installer did not see HD- yes I used the work around. I also have a newer laptop that I install 2.16 (cleaned HD before starting) Gparted - then try to upgrade to 2.16.1 (cd boot) Uni-installer did not see HD. Try agian with 2.17 (cd boot) same problem uni-installer did not see HD!!!!!!!!!! The Uni-instller has no problem with clean HD's - Only with HD with puppy installed. Tried with four desktop PC's same out come: No problem if HD is clen or has no puppy installed, IF puppy is installed does not see Hard drive!!!!!!. Please recreate problem as out line. There is a problem!!!!!!!!

Warren, I cannot reproduce your problem, which is why I am asking to look for some kind of mismatch. Right now, my laptop has a frugal install of Puppy on hda6 and a full install of Puppy on hda7, and the Universal Installer sees both.

Glenn (grholme<at>yahoo.com) 
I noticed than in puppy 2.16 & 2.17 that I could not see the Partition I wanted to install puppy to displayed in the Universal installer. My workaround is to simply use GPARTED and reformat the target partition to ext3. In other words removing traces of Puppy... Then I can see it and proceed with the install. I am not certain of the following but observation indicates that the Universal Installer will not present ext3 partition in the display. This is consistent in the 4 systems I have installed puppy on.
Another problem with the installer that occurs in 2.16 & 2.17 is when ther eis an existing GRUB in the MBR controlled by another Linux in another partition. Universal Installers process tries to update the MENU.LST in that partition and fails... I simply exit and manually invoke GRUB from the Menu and install grub to the super block of the Puppy partition.
These two problems both started from Puppy 2.16 so i suspect there is probably a slight mapping problem in the script.
These are consistent in all the installs I've done. Not show stoppers but it would be nice to see them addressed sometime.
Thank you

Warren S Carl (windowsystemcomputers<at>yahoo.com) 
Anti-Spam time out is a pain for slom typer's!!!!!!!!!
IBM T20 650meg 386 ram 30gig hd cdrom
# cat /proc/partitions
major minor #blocks name

3 0 29302560 hda
3 1 27206046 hda1
3 2 2096482 hda2
# probepart
/dev/hda1|ext2|54412082|Linux Ext2Fs
/dev/hda2|swap|4192865|Linux Swap
# fdisk -l
Disk /dev/hda: 30.0 GB, 30005821440 bytes
255 heads, 63 sectors/track, 3648 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks ID System
/dev/hda1 * 1 3387 27206046 83 Linux
/dev/hda2 3388 3648 2096482+ 82 Linux swap / Solaris
Run 'puppyinstaller' and report what partitions it recognises.
hda Hitachi HT5424030M9AT00, size 30.0 GB

It seems to be fixed. Latest pupyinstaller test version here:

Warren S Carl (windowsystemcomputers<at>yahoo.com) 
Run 'puppyinstaller' and report what partitions it recognises.
hda Hitachi HT5424030M9AT00, size 30.0 GB
Note: this was from HD boot - CD boot does not see any HD.
Shall test fix. Thank you Barry!!!

Coreutils utilities added, rsync added 

I have been experimenting with compiling T2 in Puppy (this does not mean that Puppy3 will be compiled from T2, I'm just playing). I had to add three utilities from coreutils v 2.5.1: 'od', 'fold' and 'comm'. The 'od' utility was already in Busybox, but lacks a required commandline option.

T2 also required 'rsync', so I have put v2.6.8 into Puppy. It's not too big (256K -- well, that's biggish I suppose) and probably a useful utility to have anyway.

Wireless modules and dhcpcd updated 

Tempestuous has been helping others on the forum with wireless problems, and has started a thread with some updated packages:

So far, these are rt2400, rt2500 and rt2570 wireless modules, and the dhcpcd package, updated to the very latest versions. Already there is feedback from one person that their problem was fixed.

I have put these into Puppy, so they will be in version 2.17.1.


Sorry for a long delay...not that much time here.

sadly no success with wpa2...just a first try however.

I see no option for wpa2...just wpa.

My router supports wpa,wpa2 and wpa2 auto....D-Link DI-624.

With the new net-wiz I complete a wpa connection but there seems to be a long load time or not loading pages at all. Wpa2 does not complete and the scan feature gets caught in a loop and is not able to exit or cancel the operation...just a loop.

My wifi is Intel pro wireless 3945...using ipw3945d which is always seen in the wifi wizard.

I also used the new dhcpd-3.1.0.pet

I'll give 'er a go soon again as my wife needs an open wifi to use puppy on her HP Pavillion dv8000...still have Broadcom probs there...dunno why.


USB keyboard probably fixed 

The problem of a USB keyboard not working at the choose-your-country dialog is probably fixed. I use the word 'probably' as I don't have this problem on any of my hardware, however I have deduced what is causing it.

Matt B. 
Don' stop there!

What did you deduce??

It's nothing new, a familiar problem. /proc/bus/usb/devices neds more time to get fully populated. I achieved this by moving the call to rc.country, from within rc.local0, down to just before the call to rc.network.

PDQ and CUPS printing coexisting 

Some people have reported that they cannot get CUPS to work, but PDQ does. PDQ is the printer system used in Puppy prior to version 2.17. With v2.17 Puppy went over to CUPS in the 'standard' live-CD.

It is fairly simple to get them to co-exist, so the next Puppy will have both. The main difficulty is that the symbolic link /usr/bin/lpr has to be changed depending on whether CUPS or PDQ is being used, and I have put that into the printer wizards.

Note, I'm not so sure about the situation with applications that are compiled to explicitly use CUPS. All apps in Puppy have so far been compiled without CUPS, so they fall back to the generic lpr system, however many apps have a configure-option to compile specifically for CUPS. In future I will be compiling packages with the CUPS development environment present, and I don't know whether that will break PDQ.

I feel obligated to learn how to use CUPS, but my HP laser 5 is a network printer with a RJ45 cable plugged in the back and a local IP address and I have'nt figured it out yet. PDQ had an option for IP address to print to and has worked great. I will continue to look for IP address printing, but having both would be nice.

Woodeye18, I use my Brother HL-2030 as a network printer. Try using Device URI: lpd://<IP Address>/lp1 when you set up your printer in the CUPS web interface - works for me.

THX, pkt, I'll try that

menno (menno<at>fwn.rug.nl) 
"In future I will be compiling packages with the CUPS development environment present, and I don't know whether that will break PDQ."

Why , if you donn't no what happens , lpr works .

I presume apps that use CUPS explicitly will have more options and control over printing in their print-dialog boxes.

How to contribute to the Puppy project 

I am often asked about this. During 2006 and into 2007 a lot of people have come on-board, helping in various ways. What has bcome apparent though, is that there may be expectations that do not exactly match the parameters of how I manage this project. So, I have created a web page that attempts to clarify those parameters and boundaries up-front.

For example, there was a situation earlier this year where a Forum member felt victimised by other Forum members, and a couple of people queried why did I not step in and help to sort it out.
Another example is someone who may have professional programming and project management skills and may want to contribute to the project in the form of improving the management of the project.

Whatever, I've tried to cover all bases. If you think of anything else that needs to be clarified, let me know.

Here it is: http://www.puppylinux.com/development/p ... tement.htm

Well said - thankyou for this clarification on how you run (or as some think 'don't run' ..ha,ha..) the Puppy project.. :-)

Personaly, I'm happy that someone such as yourself has actually released a Linux distro ( small or large ) that is "fun" ( and addictive ) to use as Puppy is - whatever the versions.. ;-)

I just hope that you find it in you to keep going with Puppy for many years to come.. :SURPRISED

Keep up the great work ..!!


Barry- I am sure this clarification was needed. Hope it keeps some of the un-needed questions from going your way. I can't believe you have time to manage this.... I barely find time to play with it! I've been a devotee since about 2.14 or so. You have a winner. I see for the last 7 day period on distro-watch the Puppy is in the top 10 for hits per day!! You go. And DON'T CHANGE A THING about the way you manage it. Thanks for the work. The Pup just keeps getting better.

Best wishes-

Greg in Michigan

Yay, informality! :-) The only way to get things done.

And being owned by one individual always helps...


MASONVILLE, Colo. - Zoey is a Chihuahua, but when a rattlesnake lunged at her owners' 1-year-old grandson, she was a real bulldog.

Booker West was splashing his hands in a birdbath in his grandparents' northern Colorado back yard when the snake slithered up to the toddler, rattled and struck. Five-pound Zoey jumped in the way and took the bites.

"She got in between Booker and the snake, and that's when I heard her yipe," said Monty Long, the boy's grandfather.

The dog required treatment and for a time it appeared she might not survive. Now she prances about.

"These little bitty dogs, they just don't really get credit," Booker's grandma Denise Long told the Loveland Daily Reporter-Herald.

Very true, the little bitty dogs just don't get no respect!

I would have expected everything on that page to be obvious to any person following this blog..

Anyway, a couple of little things:
1) although I personally can't be bothered with bug-trackers (even got Benny from XFCE pissed off by not using Bugzilla…), I think using one (and having a "report bugs" link on the front page of puppylinux.com) could have cut down the harassment from users.

2) (might be considered petty) having the odd-numbered versions as the unstable ones might not be the best option: major upgrades tend to use 2.20, 3.00 etc… and they usually are less stable than their successors (thus a lot of stable packages have versions like 3.01 -- bugfix releases after a major upgrade).

Ted Dog 
reply to Dougal, That would be the Microsoft way, rush crap out the door make it new and improved marketing then release fixed versions. Think about linux kernel odd numbers like 2.5 = new radical thinking, good stable stuff moved up to 2.6. It sort of an inside 'joke' how many good stuff remain in the linux world make it to version one.O wine for example.

Ted Dog 
yep still sub 1.0 releases first beta last year (a decade after project started)

Ted Dog, I don't have any MS products. All I am talking about is GPL sruff that compiles on Linux that I see on Sourceforge or such.

An example from the Busybox changelog entry for 1.6.0:
Since this is a x.x.0 release, it probably does not deserve "stable" label. Please help making 1.6.1 stable by testing 1.6.0.

Bob Bagwill 
Barry, don't stop having fun! That's an order!

To take the heat off Barry, how about if everyone goes back to using Mantis, http://tracker.nfshost.com/mantis , which is apparently still going strong?

USB Flash: flushing RAM. Plan for 2.18 

Booting from USB or IDE Flash drive, Puppy 2.16 had flushing of RAM to the pup_save file, rather than copy-down in earlier puppies. Puppy runs in RAM as much as possible to avoid writes to the Flash drive (lots of writes means lower life-span for the drive), so for example if you install some packages they will all be in RAM, but are flushed to the pup_save file on the Flash drive on a batch basis, normally every 30 minutes (or when RAM gets full, or you can do it manually).

However, I have disabled this in 2.17, gone back to copy-down (so RAM doesn't get flushed until you reboot), as there was a problem with flushed files potentially getting corrupted.

If you install big packages and run out of RAM, it is a nuisance to have to reboot, so we would really like to have flushing back. The good news is that forum member Andrei has been working on safe flushing -- basically by restricting flushing to just-installed packages. In other words, you install a PET package, it gets flushed immediately and RAM free space is not reduced. We have to sort out some details, like integration with PETget, and do some testing, but probably expect this in 2.18.

Note, I intend Puppy 2.18 to be basically 2.17 with small refinements, aiming at cleaning up a few things, fix the Universal Installer, just a general tidy-up. With that in mind, as there won't be any major changes, will probably bring out 2.18-release-candidate then the final. I can't estimate when 2.18 wil be out, but probably much less than our usual 6 - 8 week cycle.

Another thing I was thinking is we could adopt odd numbers as more adventurous versions, even numbers as refinements of the previous odd-numbered version.

I think your last comment is an excellent idea. It allows us to point newbies and non-developers (regular users) at the more stable versions. It also focuses work on cleanup of small flaws in those versions, that tend to get ignored when bigger stuff is going on.

Sounds good - as you will remember, golden rule: never buy M$DOS with an even number! 4.01 with Edlin was a milestone. Sorry, my typing isn't too good - make that millstone!

Helloe Barry

If you are working on the Universal Installer, have you thought of doing it as a tabbed dialog/wizard, where each step in the process progresses you through the tabs? I find the series of popups in different parts of the screen disconcerting and I think somebody new could have trouble figuring out what's going on (they can't go back to check).

The same principle could apply to some other parts of puppy such as the Connect Wizard whihc is also a series of popups.

I think Marks GTKBasic is capable of doing these sorts of dialogs without much effort, I would be willing to try the UI stuff myself as long as the current under-the-hood code could be utilised, not for 2.18 though I'm pretty busy at the moment.


Mr Doolie 
Speaking of the universal installer, there's a bug thats been there since way way earlier. When installing to a flashdrive or external drive when the installer says "Making xxxx bootable" it's lying. I've always had to make the device bootable by hand using cfdisk.

EZpup: give your pup a makeover 

WhoDo has been pioneering an interesting approach to giving Puppy all the "bells and whistles" and "shiny bits" not in the standard release. He has created a PET package (15MB!) that you just download then click on to install (and uninstall using the PETget package manager if you don't want it anymore). WhoDo was the team leader for the 2.15-Community-Edition, but then he decided why not just add all the extra goodies as a PET package to the official Puppy? -- hence EZpup.

EZpup has just been upgraded to suit Puppy 2.17. Read all about it in these forum threads:

WhoDo has also created an introductory wiki page:

Eric/caneri (ericmulcaster<at>gmail.com) 
Hi Barry and crew,

I have 2.17 with EZpup installed on a Verbatim 2gb flash....

This is stellar!!!!...I have fallen in love with the pup all over again.

My Puppy Software Installer does not populate the list of packages though...just says Pet Packages-2 and Pupget packages-1 and when I try to go to some package it says I need to select a package..but there it was gone...no packages to choose from also the same with the search feature.

I know I haven't done my homework yet but your new puppy is a work of art.


Eric, please try this updated version, it now uses a one minute timeout instead of six seconds:
updated PSI


Mark and all the puppies,

As usual your skill has been stellar!!!!

Thank you very much for your post and update...it works flawless.

I can see a day when your work will be recognised as a nexus for linux.


Very nice and still under 100Mb - could be less? Particularly liked the bulldog background - everyone should have access to this one, Barry. Some difficulties occur swapping desktops?
Don't usually go for the aesthetics, but this one is different. Perhaps this is a good time to swap Opera for Seamonkey before the final emerges?

F M Lynch (fmlynch2<at>comcast.net) 
Prior to installing EZpup, My frugal HD installation of Puppy 2.17 automatically detected, configured, and activated my Prism2_usb module on bootup flawlessly. After installing EZpup my Prism2_usb module was not detected. I also lost some of my Open Office desktop icons, much to my dismay. After uninstalling EZpup, my Prism2_usb module was still not detected, and all my Open Office desktop icons were gone.

I had fortunately saved a backup copy of my Pupsave_2sfs file which restored everything back to normal.

[i]Eric, please try this updated version, it now uses a one minute timeout instead of six seconds:[/i]

Thanks, Mark. Future versions of EZpup will include the revised timeout settings.

[blockquote]After installing EZpup my Prism2_usb module was not detected. I also lost some of my Open Office desktop icons, much to my dismay. After uninstalling EZpup, my Prism2_usb module was still not detected, and all my Open Office desktop icons were gone.[/blockquote]

EZpup does not impact on any modules. It is a look-and-feel update only, with the one exception being an improved xcompmgr. One possible cause of your problem might be that your pup_save file was somehow corrupted? EZpup takes quite a while to install, and if it is interupted before it finishes you can get some very "uneven" results.

F.M., that was wise making a backup copy of pup_save.2fs -- advice to all newcomers, it's easy to do, just go to /mnt/home and right-click, choose to make a copy. A good precaution to take before doing any major surgery.

Oh yes, name it something like this: pup_save-2.2fs
Puppy will recognise a name like that at bootup and offer load either one, your choice. Or, if your experiment went wrong and you want to dump pup_save.2fs and go back to the backup, well, there are various options, you could just bootup and choose the backup, then delete the unused file, or bootup in ram with 'pfix=ram' and restore the backup to the original filename, then reboot.

Two bugs in 2.17 

It seems that some naughty USB keyboards are still giving trouble, even though my 2.17-release-candidate-testers reported the problem fixed.
It is a little once-only hiccup when you boot Puppy for the first time, or without a 'pup_save' file by using the 'pfix=ram' boot parameter. A dialog window comes up asking to choose your country, and some people have found their USB keyboard does not work at that point, but once past the dialog it does. You need to plugin a PS/2 keyboard to get past that little hurdle.
If you have a computer with only USB keyboard no PS/2 (which is very few people. All laptop internal keyboards are PS/2), most likely you won't hit this bug as it is fixed for most USB keyboards. If you do fall into that unlikely category, please wait for 2.17.1, as I know how to finally fix it.

Some problems are being experienced with the Universal Installer not recognising all hard drive partitions, at least some people get this problem -- the Installer works fine on all my PCs. Anyway, one person has reported that an older version of the Installer works. This bug started with Puppy 2.16, it is not specific to 2.17. The Universal Installer had a major overhaul for 2.16, see the announcement in this blog:
http://www.puppylinux.com/news/comments ... 507-075511

Perhaps the Installer from Puppy 2.14 will work for those who have this bug. Forum member RazzBear says this is the case, and has posted an older 'puppyinstaller' here:
You can try that if you like, copy it to /usr/sbin, probably name it something different from the original file. Let me know how it goes.

In the meantime, I'll study the two puppyinstaller scripts and see what I can come up with.

Please note, for most people the Universal Installer will work fine. I notice that those who have reported missing partitions all have hard drives with a dozen or so partitions, which could be a factor.
The two bugs reported here will not occur for the vast majority of people, but just in case you hit one, the above notes will help you to work around it.

You might also look at the mod I suggested at the end of the 217B thread for inclusion in 2.17.1:


This is a reworked usbstoragecompletionfunc that should alleviate a hang during boot I have run into (there were two loops in the previous version of that function, only one of which was timed out).

John Doe has made a new initrd.gz for me to test it. I couldn't make one myself; for some reason the edit-initrd thingy is broken, or maybe I used it wrong!

hda1 vfat w. prestine puppy
hda2 ext3 w. full install puppy
hda3 swap
Universal installer in 2.16 vorked O.K. 2.17 did just see hda1.
So could it bee that the seek for puppy stoppet when the first is found ?

John Doe 
> for some reason the edit-initrd thingy is broken, or maybe I used it wrong!

i made it wrong :-(

we'll get that worked out and tested barry, you stick with the USB keyboard and universal installer error.

I have posted a puppyinstaller script here for testing:

Have Puppy 2.16 installed on old Dell GX1 350Mhz, 64M Ram (plus swap), frugal install along side with W98SE --- works A-OK. Systems has a hardware modem, but also has an Intel537 in slot.

The Upgrade to 2.17 worked OK --- got the 2.17 desktop OK.

However on reboot system crashed with this error message:
Modules linkd in: Intel537(P) 3c59x mii i2c_piixt

...etc... Call trace ...etc...lots of the technical type stuff you'd expect to see but I can't simply cut and past so these few lines will have to serve as a hint....the last couple of lines of error drump were...

Code: Bad EIP value
EIP: [<00000000>] rest_init+0x3feff000/0x20 SS:ESP 0068:c39a5f78

I've tried to comples the drivers for the Intel modem myself and has always ended in failure (that's why I spent $5US to get a hardware modem to play with Linux with).

Puppy 2.17 loaded and worked A-OK on all the other old PCs I've tried it on (even works well on very old HP PII 233Mhz, 96MB RAM+Swap!).

Please keep up the good work --- don't let those few Bozos, those technical guru wannabies destract you from your Linux Quest. Thanks.

I don't know if this is related or not, but I can't even enter the "puppy pfix=ram" boot parameter with my USB keyboard. I have to plug it into the PS/2 port (using an adapter, of course.) The USB keyboard works fine after Puppy has fully booted.

Hi Barry,

Puppy 2.17 works great on 4 machines that I've tested (2 notebooks, 2 pc), but did not start at a friends pc

Motherboard: MICRO-STAR MS-7091
CPU: Intel(R) Pentium(R) 4
BIOS: Phoenix - AwardBIOS v6.00PG
2 Harddisks: both Western Digital Corporation WD1600JD-00HBB0
sda1: NTFS A0B4-263C 76340 MB
sda5: NTFS 7051-0E81 69288 MB
sda6 FAT32 3457-7E04 6983 MB
sdb5: NTFS F8E7-9E4E 152617 MB
hda: Pioneer DVD RW DVR-109
MEDION RADEON X740XL Secondary (PCI\VEN_1002&DEV_5E6B&SUBSYS_17811462)

Booting just stopped right after the message:
creating unionfs on (/initrd)/pup_new (to become '/')... done
or sometimes even one line before

I also could not get Puppy 2.17 to run on VirtualPC.

Puppy 2.16 however works fine on all machines.

What could be the reason for that?


Nice Job!

This looks like a great release, but I did have a scare....

I must have tried to install a corrupt copy from a Flash drive,
and it caused the Puppy Package Manager to go blank- both columns!
Traced it to a bad copy of qucs. Re-installed 2.17, deleted the bad
file, and installed a good one, all is well. The copy of
qucs-0.0.12-i486.pet that I posted on the forum was clean---phew!

Thanks Guys- 2.17 is a Winner!

rxr, the Intel537 module causing a crash ...why doesn't that surprise me... it will be interesting to get experiences from others on that modem chip.

Still a Bug w/ Synaptic Mouse Pad Not Working on M' Toshiba Satyrlite M50 Laptop :-(

tysee (ebbe<at>m0m.se) 
I have had trouble with USB keyboard.
It disapeared when I activated legacy support for USB.
Keep up the very good work.

Bob Jackson 
tightvncserver is acting strangely under 2.17

The server starts OK, but when i connect with a vnc
client (either locally or remotely) all I see is the
toolbar at the bottom of the screen. I don't see
any desktop icons.


marcio js (sofmodemintel537<at>yahoo.com.br) 
> BarryK
> Saturday, July 21, 2007, 09:03 PM
> rxr, the Intel537 module causing a crash ...why doesn't
> that surprise me... it will be interesting to get
> experiences from others on that modem chip.

Well, I find that we have the same problem ...

If I use Puppy 2.17 whitout saving personal data/file, it works fine and I can reboot witout crash...

asus a7s266-vm, AMD Durom 1300MHz, 256M RAM, HD Samsung 20G
and SofModem Intel537 v.92 (Windows say: data, fax and voice modem)

Call Trace:
[<cecc97b7>] ham_proc_shutdown+0x11/0x35 [Intel537]
[<cecc6073>] get_pci_info_537+0x73/0x3da [Intel537]
[<cecc6a19>] core_cleanup_module+0x5/0x10[Intel537]
[<ce806018>] init_serial+0x18/0x7b [Intel537]
[<c0132f22>] sys_init_module+0xe2/0x150
[<c0103ea4>] sysenter_past_esp+0x5d/0x89

Code: Bad EIP value.
EIP: [<00000000>] rest_init+0x3feff000/0x20 SS:ESP 0068:cd5fbf78

Puppy is "pure Heaven" 

Find out why:

Eric Mulcaster 
Just so happens I delivered, what had been an overheated broken Dell slimline back to a lady last night. It now has 2.16 loaded on a 1gb sandisk cruzer flash drive sticking out of the back of her computer.

She had a bad HDD and mobo appeared to be dead......Dell support said it was not repairable.(HD and controllers I guess). Now running with no hard drive at all and 1/2 gb ram. Hooray!!!

This saved her lots of money and her computer is not in a landfill somewhere. She was amazed at the speed of Puppy and that a virus scanner wasn't really needed to go online. Needless to say she loves her new puppy and the exploding sheep!

Well done Barry and crew...I already use a usb install so I knew she would be pleased.


Eric Mulcaster 
I should add to the above post.

The Live Earth people and Al Gore the "Goracle" would like the idea of the pup and it's potential to save resources.


Congratulations on the well-deserved good press. Puppy is really starting to come into its own as a very nice, simple OS for new and power users. Keep up the great work. I think I speak for the silent or at least quieter majority when I say that your work with Puppy is greatly appreciated and let us know when we can help. I have used Puppy for nearly 2 years now and I keep showing people the amazing things it can do in such a small space. Puppy has been the central focus of my learning about using Linux. Thanks!

nice article :-)
Gizmo should install PSI to get easy access to more packages.
Does someone know where to get the background from the DSL screenshot ?
Is it part of DSL?
Could someone upload it to the forum please?
Thanks, Mark

ok, mounted a DSL-image using a selfcompiled cloop module.
Here is the picture:


Ted Dog 
What only 15 pages to get WindowsXP in a flash drive, with no help from Bill and his well paid crew. When a man in the middle of no where, still using and supporting dailup does it for free and his support is free. Yes its almost like heaven. With a cheap 256M pen drive and a simple remaster script I can boot five very different machines and it remembers the Xvideo settings for each very nice!

(c) Copyright Barry Kauler 2008. All rights reserved. http://puppylinux.com