Puppy developer news:
| Older news||Later news
May 18, 2006
Another one bites the dust! Jesse Liley has released MUT version 0.0.8, which has some fixes to
work on Puppy2.
Lior Tubi has created a great little enhancement to the clock in JWM,
Clicking on the clock now brings up a calendar. I think someone mentioned
awhile ago on the forum that this would be a good thing, as they could do
that in Windows and it is very convenient. Thanks to Lior we have that
convenience now also!
The kernel has been recompiled with FUSE support built-in (not as a module)
and with 4K stacks disabled (using 8K stacks, same as the k2.4).
2.5.3 has been compiled and the library libfuse is now in puppy2. This is
required for ntfsmount, the program to mount NTFS partitions such that files
can be created.
Puppy2 now has the utility applications mkntfs, ntfsmount and ntfsresize from ntfsprogs version
2.5.3. Ntfsmount requires libfuse from the FUSE package.
Note, the kernel also has a NTFS driver, which I chose to be built-in.
This is somewhat less mature but can safely use a pre-existing pup_save.3fs
file. As it improves, it will eventually make ntfsmount unecessary.
USB mouse problem solved. I found a computer that exhibited the bug, and
found the solution.
If you are testing the April28 alpha snapshot, you should be able to
verify by exiting from X then remove modules:
# rmmod usb-storage
# modprobe uhci-hcd (or ohci-hcd)
That last line should show any USB mouse and keyboard. The mouse should
also work when X is started.
Well, it seems something has gone wrong with USB mouse detection in the
move to the 2.6 kernel. Heaps of bug reports for the 28th April snapshot "my
USB mouse doesn't work!".
There also seems to be a problem with writing to NTFS partitions. It is
supposed to be enabled. It is enabled in the kernel, but it seems that just
using "mount -t ntfs /dev/hda1 /mnt/data" does not allow file creation. I
skimmed through the linux-nfts.org website too quickly, thought the driver
now supports file creation, however on closer examination, the ntfsmount
program is needed -- okay, it will be in the next snapshot.
http://www.kcore.org/?menumain=4&menusub=2 has this:
Hardware: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) AC'97 Modem Controller - Winmodem.
This modem can be gotten to work using the Linuxant HSF Softmodem drivers.
Unfortunately, they are payware.
NOTE: You have to compile your kernel without CONFIG_4KSTACKS! If you use this driver with
4K stacks enabled, it _will_ crash your system!
It seems to be another case where the kernel developers are trying to improve performance, like the 100Hz versus 1000Hz interrupt, but it causes compatibility problems.
The option CONFIG_4KSTACKS isn't even in the .config file, but I found that the "Kernel hacking" option has to be turned on, then it is offered as a sub-option. It's default is "y".
So, if you test ndiswrapper, it may crash the kernel, with symptoms like frozen keyboard and mouse.
I forgot to include the ndiswrapper kernel driver in the latest puppy2
alpha. So, I have compiled ndiswrapper version 1.15. The Unleashed package
with the executables and the kernel driver module are downloadable from
You will need to gunzip the module, install it to
/lib/modules/18.104.22.168/misc/, then execute depmod. The executables can be
installed using PupGet or manually.
Note that the ndiswrapper package warns that the kernel is compiled with
4K stacks, which may cause some Windows drivers to crash. I need to recompile
with 8K stacks, which will do for the next release of Puppy2.
Note also, I compiled ndiswrapper running Puppy live-CD, but I had to add
some extra files to devx_003.sfs, in /usr/src/linux-22.214.171.124/. An updated
devx_xxx.sfs file will be uploaded soon.
Ted Dog has kindly mirrored the latest Puppy2 alpha 28th April 2006 here
Puppy2 alpha-28apr06 is released. This is not an official release, it is
for Puppy testers.
Download the complete set of modules for the 126.96.36.199 kernel, also the pup_003.sfs compiler file: http://www.puppylinux.com/test/
There is a special forum for puppy2 development. Report here any bugs, I also want to know about success stories!
Forum URL: http://www.puppylinux.com/nfphpbb/
Puppy2 discussion also takes place on the regular forum: http://www.murga.org/~puppy
It is intended that the next release will be puppy2-beta, and this is the to-do shortlist:
JustGreg suggested a small improvement to wvdialshell script, that I have
The ALSA Sound Wizard, which is a
modified alsaconf script, runs at first bootup. I have modified the script so
that if executed with the parameter like this: "alsaconf --auto", then no
questions are asked. Previously, the script asked many questions, now at
first bootup it just runs and you get two barks if successful -- this is
great for users who often want to start Puppy running in ramdisk only, no
prior saved configuration.
I had a bit of a play, created a little taskbar applet that displays free
memory. Regardless whether you have a pup_save.3fs persistent storage file,
have Puppy installed to hard drive, or running totally in ramdisk, this
applet displays a single value that tells you how many megabytes you have
left. It displays as white text on a dark-blue background, which changes to
dark red if memory drops below 20M, then flashes lightred-darkred if memory
falls below 5M.
I was concerned about the new block of x_tables modules and some older ip modules missing, however it appears all
is well. One of the new modules is 'xt_state.ko', and the apparent old
equivalent that is missing is 'ipt_state.ko'. However, when I run "modinfo
xt_state" to get information about the module, it is reported to have the
alias name of 'ipt_state'.
I have recompiled the 188.8.131.52 kernel with CONFIG_DVB=y, and all the
drivers for digital video broadcasting as modules.
Lior Tubi's wonderful Blinky
network monitor is now upgraded to version 0.8.
As a temporary measure, I have restored the v6.8.2 i810 Xorg driver in
Puppy2, as the one in v6.8.1 does not work very well. The April 11th snapshot
has the v6.8.1 i810 driver, which I had gone back to as the v6.8.2 i810
driver has a resource leak.
Upgraded Mark Ulrich's patched Dillo v0.8.5 to the April 24th release.
The iptables/netfilter kernel modules in 184.108.40.206 kernel have changed from
earlier 2.4 and 2.6 kernels. Some 'CONFIG_IP_NF_' configure options are
missing, but there are a whole new bunch starting with
'CONFIG_NETFILTER_XT_', which generate a new bunch of modules, all starting
with the name 'xt_', for example 'xt_conntrack.ko'. So, I've put many of the
old modules that still exist into Puppy, and added all of the 'xt_' modules.
Just guessing here, seems like what is needed.
I have updated dvw86's Trashcan
to version 2beta. Note for developers, the trashcan is located in the
rox_filer-1.2.2patched Unleashed package.
Joe's marvelous window manager JWM has been updated to version 1.6. The
whole thing is still a single 125K executable.
The Ethernet/Network Wizard has
been upgraded by rarsa to version 23.
I have compiled Transmission version
0.5, a small and very easy-to-use BitTorrent client. This is now an Unleashed
package and will be in the next snapshot of Puppy2.
Script /etc/rc.d/rc.update has been fixed to work in Puppy2. At least I
think it has, not tested yet. This means that if you boot a new version of
Puppy2 and there is already an existing personal storage file (pup_save.3fs
or folders from a multisession-CD/DVD), then all critical system files are
ensured to be updated.
The 220.127.116.11 kernel supports safe writing to NTFS partitions (according to
the documentation on the Linux-ntfs website), so Puppy2 now has full NTFS
write-support enabled. This removes all the earlier restrictions, meaning
that a personal storage file (pup_save.3fs) can be created on an NTFS
partition just like any other. Also, Puppy can be installed to a NTFS
partition just like previously possible to a MSDOS or VFAT partition(by
option-1, also known as the "poor mans" install mode -- the Puppy Universal
Installer will explain that!).
The Puppy Universal Installer needs a bit more work due to the move to the
2.6 kernel. The problem is that 'tiny.exe' from the Gujin package is used to
create a boot floppy, and previously the Linux kernel 'vmlinuz' was also
placed on the floppy. However, the kernel is now 1618K, too big for the
floppy, meaning that it now has to be on some other FAT partition. The
restriction to FAT partition is because tiny.exe is a MSDOS application and
FreeDOS is used, which only recognises FAT partitions (msdos or vfat). The
same thing goes if use Syslinux.
The following modules have been added to Puppy2:
net/tulip/uli526x.ko ULi M5261/M5263 fast ethernet driver
VPN developers/users, note that Puppy 1.0.8 has ppp_mppe_mppc.o driver,
but a MPPE driver is now in the 2.6 stock kernel, called mppe.ko. All of the
above are in the stock (original) kernel.
Bug update: I think that I have fixed the problem in Puppy2, when bootup
and it stops at the prompt, X doesn't start.
I've worked out a solution. Puppy needs to detect whether a PC has a USB
interface in the init script and also later in the rc.sysinit and rc.modules
scripts. This means that lspci and pci.ids have to to duplicated in both. The
lspci executable is very small (33K), however the latest pci.ids file is 380K
-- however, I was able to extract just the USB information, so in the initrd
the pci.ids file is only 5.5K. So, it's not much overhead for the initial
I did mention that there are some "issues" with the new 18.104.22.168 kernel.
One of those is that the file /proc/pci has different information about the
PCI interfaces, compared with the 2.4 kernel. It used to have the
description-string of each interface, which is now replaced by a line that
commences with "Class ..." followed by various numbers, including the
vendor:chip ids. It was the description-string that I have used for various
purposes, such as determining if the motherboard has usb capability.
USB Controller: Silicon Integrated Systems [SiS] USB 2.0 Controller (rev 0).
But, running the 22.214.171.124 kernel, the above line is replaced by:
Class 0c03: PCI device 1039:7002 (rev 0).
This change means that a vendor:chip id database is required to lookup
what each interface does.
The lspci utility does not help, as it only returns the same information.
It can be made to return more info if the pci.ids database file is present --
which it isn't. The problem also, even if Puppy has the pci.ids file, it is
always playing catchup. I found some information on the Internet, it seems
the kernel developers dropped the description information in favour of using
lspci -- but lspci does not read the descriptions directly and relies on the
What about the kernel log during bootup, as returned by dmesg? No, it
seems not, no interface descriptions until the USB drivers load.
What I have tested, and really I hate doing it this way, is try to load
each module in turn: usbcore, uhci-hcd, ohci-hcd and ehci-hcd, trusting that
it will fail if the appropriate hardware is not present. No, on my PC that
only has a USB1 interface, the USB2 ehci-hcd driver loads without
Bashdiff is an
enhancement for Bash. Truly great additions, too good to resist! I've created
an Unleashed package, bash_bashdiff-3.0-1.33,
which is Bash with the Bashdiff patch. This installs 'bash' and 'bashbug'
into /bin and 'libwilliam.so' into /usr/lib, as well as a help file
'bashdiff.htm' into /usr/share/doc.
These packages have been upgraded for the 2.6 kernel: module_init_tools-3.2.2, e2fsprogs-1.38, wireless_tools-28 and unionfs-20060417 (cvs snapshot). The PCMCIA package remains at the same version, but recompiled for the 2.6 kernel: pcmcia_cs-3.2.8 -- this package may be dropped sometime, due to equivalent functionality in the 2.6 kernel.
Note also, the 126.96.36.199 kernel was compiled from almost-pristine source,
with just one patch, squashfs-3.0.
I have created a Unleashed package of Mark Ulrich's brilliant new desktop
background setter, Puppybackground. This
will be in the next Puppy2 snapshot.
I'm typing this from the latest Puppy2 snapshot, running kernel version
188.8.131.52. Some small issues, should be solvable. What prompted me to do this
is the unionfs we have used up to now is version 1.0.4, as that is the last
version that runs with the 2.4 kernel. Development of unionfs has continued
for the 2.6 kernel only. The 1.0.4 version has a bug, in which files and
directories are sometimes invisible -- this bug affects both Puppy1 and
Puppy2. I went right to the bleeding edge, grabbed yesterdays cvs snapshot of
There's lots of good stuff in the 2.6 kernel. I shall be experimenting to
see if NTFS write support is now solid.
Puppy2 now has working automatic mouse detection, at least on my computer!
Plug in anything, USB, PS/2 or serial, boot Puppy, it works. When Puppy is
running you can still use the Mouse Wizard, which is handy if you have two
different types of mice connected -- the Mouse Wizard lets you choose which
one you want to use, whereas the autodetection will select in order of USB,
serial, or PS/2 (meaning that the USB mouse has highest preference).
I have made an Unleashed package, dillo-0.8.5patched, which is based on a
patched version of Dillo v0.8.5 from http://teki.jpn.ph/pc/software/index-e.shtml
and further adapted by Mark Ulrich to be compatible with his
PB-DebianInstaller program (still in alpha, not officially announced yet).
Okay, here it is!!! puppyserialdetect, a
little program to detect serial mouse and serial modem. Download it from
# ./puppyserialdetect 2>/dev/null
to make sure that stdout returns nothing.
There is a thread in the main forum discussing this, please read before
trying puppyserialdetect: http://www.murga.org/~puppy/viewtopic.php?t=7475
A clarification: I stated below that the detect library code cannot detect my USB
mouse due to a difference in /proc structure. Precifically,
/proc/bus/input/devices file is missing, as this is only available with the
I seem to have basic detection of the serial mouse working. Despite
extremely rusty knowledge of C, I managed to extract what I needed from the
detect package, which is a hardware
detection library. From the copyright notices in the code, they in turn got
some of it from RedHat and Mandrake.
There is some discussion on the forum about the awkwardness when you want
to use different persistent save files (pupxxx in puppy1, pup_save.3fs in
Note that when you boot off a USB or hard drive partition, Puppy always
looks in that (boot) partition for the pup_save*.3fs file(s). Puppy will not
wander far and wide over your PC and discover some stray pup_save*.3fs
I seem to have misunderstood how the isolinux.cfg file works. At the
Puppy2 boot menu, typing 'puppy acpi=on' does not work. The isolinux.cfg
The total effect is that we get this boot commandline:
root=/dev/ram0 initrd=initrd.gz acpi=off PMEDIA=idecd BOOT_IMAGE=vmlinuz acpi=on
...and it seems only the first occurrence of 'acpi=' takes effect.
There has been a teensy bit of reorganising of the main menu.
I've expanded the live-CD boot menu. As well as the usual kernel options
like 'acpi=on' or 'ide=nodma', you will now be able to choose to ignore any
existing saved sessions and run totally in RAM, or in the case of
multisession CD/DVD you will be able to choose to ignore any number of
most-recent saved sessions.
But then I realised it is only half-done. What we really need is a
mechanism to mark folders permanently as "bad". So, I implemented that too.
If you roll back say by 2 folders, they are automatically marked as bad and
on future boots will be jumped over.
In Puppy2 first boot, the session can be saved to floppy disk. However,
this causes trouble on PCs that do not have a internal floppy drive. The
floppy detection code at startup and shutdown has a long delay on such PCs,
and we have not come up with a reliable alternative. However, the bottom line
here is that you have to be desperate to save the session to floppy disk
anyway -- it only has room for the basic configuration files. So, the simple
solution is to restrict saving to floppy disk to non-USB computers. Older
non-USB computers will all have floppy drives, except for some rare ones. So,
on all USB-enabled PCs, the check-floppy delay is eliminated.
I've restored a default index.html page in /root/ghttpd/htdocs/.
A Puppy2 tester reported that dragging a file to pupzip, the desktop
drag-and-drop archiver, failed to create an archive. I tracked this down. One
of the scripts used by Xarchive, the archiver program called by pupzip, in
turn calls the utility program mktemp. Now, mktemp is a Busybox applet, and,
as so many times before, it lacks the required commandline option ( -t ). So,
I got the full mktemp out of Vector, and now pupzip works fine.
In response to a request on the Forum, I have improved the man script. Previously, typing "man
something" did not always find the help file even if it existed. This was due
to some help files being named 'index.html' or something other than the name
of the program, and to a limitation in find itself -- it will not follow
paths with symbolic links, or rather can only be made to with some
jiggery-pokery code that I had to hack together. I have fixed this, also if
any Busybox applet is specified then the Busybox help page will open -- for
example, "man zcat".
Of course, if online, if the help is not found locally then the help
system could fall back to an online man pages site -- anyone know a good
ad-free site with all the man pages in html?
I was replying to a discussion on the Forum about saving the current
session to a new multisession CD/DVD, then realised something was not quite
right in the session-save code in Puppy2. They were discussing Puppy1, but I
responded that it is very easy in Puppy2, but then realised the problem. Now
A clarification here. A "session" refers to all the new and changed files
since you booted up. However, in the case of moving to a new CD/DVD, the
accumulated sessions have to be burnt to the new CD/DVD -- email, browser
history, application configurations, personal files, everything. So, hitting
the "Save" button puts all of that on the new CD/DVD (with the exception of
old, deleted and archived files, that are left behind on the previous
Technical note: the scripts involved are /etc/rc.d/rc.shutdown and
I still haven't properly resolved automatic mouse
detection. There is a possibility, for someone with C programming skills, who
is also interested in the hardware side of things.
Ted Dog has a fast download site for the latest Puppy2:
A snapshot of Puppy2 alpha, dated April 11th, has been uploaded. This is
not an official release. It is for Puppy testers and developers.
Forum URL: http://www.puppylinux.com/nfphpbb/
Puppy2 discussion also takes place on the regular forum: http://www.murga.org/~puppy
The utility install has been placed
in Puppy2. This is from the fileutils-4.0 package. It is in usr_devx.sfs (or
devx_xxx.sfs in Puppy2) file, but it is needed in the standard live-CD.
Xorg has a dozen or so hardware-specific drivers, but they do not always
work well. The Intel 'i810' driver for example. Or there may not even be a
driver to match the video hardware. But, Xorg has a generic driver called
'vesa', which works kind of like the Xvesa Kdrive X server (but do not
confuse the two!), and should work on most video hardware, but with some loss
of features and performance compared with a hardware-specific driver.
Mark Ulrich (MU on the Forum) now has a PayPal button at http://dotpups.de/. I've donated US $62, just a
token to say thanks for his input (just redistributing some of what others
have donated to me!).
The Xorg Wizard had a problem, when the test X window is displayed there
is an "okay" button, but it was only selectable by the mouse. If the mouse
isn't working then the button could not be clicked. Now it can be selected by
the ENTER key also.
The Xorg Wizard now has a dialog to choose the mouse, in case the
autodetection at bootup got it wrong.
I have restored the correct i810_drv.o driver file in Xorg version 6.8.1.
This is the version currently used in Puppy. The i810 driver had some
problems, and we substituted the i810 driver from Xorg v6.8.2, however it
appears to cause a resource leak.
Xvidtune is a GUI utility to adjust the screen up/down/left/right and
width/height, in case the current Xorg mode is not rendering quite right on
the monitor. I have now incorporated Xvidtune into the Xorg Video Wizard, so
that changes created by Xvidtune can be automatically inserted into the Xorg
configuration file /etc/X11/xorg.conf.
Program fc-cache is added to
Puppy2, from the Xorg v6.8.1 package.
Pakt has improved the xkb directory used in the Puppy2 snapshot of March
29, adding more languages and cutting the size down. This (improved directory
/etc/X11/xkb) is now in Puppy2.
There are reports that the md5sum is wrong for the 29th March Puppy2
snapshot. So, it is also uploaded here:
A snapshot of Puppy2 alpha, dated March 29th, has been uploaded. This is
not an official release. It is for Puppy testers and developers. Get it from
Forum URL: http://www.puppylinux.com/nfphpbb/
Fixed some more bugs in init and rc.shutdown scripts.
I have decided to release the puppy2 alpha snapshot later today, as have
to do something else (non-Puppy) for the next few days.
I haven't completed chasing down one bug. The first time a CD/DVD is
shutdown, the session can be saved to floppy disk. That part works, but at
bootup there is a problem. The saved session, pup_save.2fs, is found on the
floppy and mounted, but then something goes wrong. I have to leave that one
for now, as have to do something else, but if the idea of saving the session
to floppy interests someone, feel free to investigate this bug!
Okay, I fixed a bug in rc.shutdown, and the full hard drive installation
now shuts down properly.
I'm testing a serial mouse in Puppy2, it sure is "nervous". Single clicks
are too easily interpreted as double clicks. In ~/.jwmrc and
~/.jwm/jwmrc-personal, there is a setting for double-click speed -- I made
that a bit bigger, but not sure if it has improved. If this is a JWM problem,
then the same symptom will be in 1.0.8r1. Anyone with a serial mouse who has
this problem? Any thoughts on fixing it?
<!-- Double click speed (in milliseconds) -->
In Puppy 1.0.8r1, xkb is somewhat broken. This is the keyboard layout
system for Xorg, and the directory in question is /etc/X11/xkb. Puppy2 alpha
snapshot of 12 March 2006 has the same xkb folder as 1.0.8r1.
JustGreg and I are chasing down the bugs in the "option 2" (full) hard
drive installation, using the Puppy Universal Installer in Puppy2. I found a
bug in the /usr/sbin/grubconfig script. After fixing that, I successfully
installed Puppy to a ext3 partition and created both a boot floppy and
installed Grub into the MBR.
My Pmount drive mounter program for puppy2 now handles Zip and LS120
internal ATAPI (IDE) drives.
The Puppy Universal Installer for puppy2 now installs to Zip and LS120/250
JustGreg did some incredible research into getting USB Flash drives to
boot puppy2. Forum thread:
I'm doing some more work on the Puppy Universal Installer for Puppy2. The
"option 1" (poor man's) hard drive install has been improved.
If you are testing the March 12th snapshot of Puppy2, multisession CD/DVD,
don't bother with testing what happens when the CD/DVD gets full. At
shutdown, or whenever a session is saved, you will be asked to insert a new
blank CD/DVD, but the old session wasn't properly saved to the new CD/DVD.
I've just fixed it.
The new CD remaster program for Puppy2, written from scratch, is now
working. Nice GUI interface, easy to use. The script is
I'm writing a remaster-CD script for Puppy2. Getting there!
JWM is fixed, plus Lior is adding more features! See thread: http://www.murga.org/~puppy/viewtopic.php?t=6907
Mitch Bradley has studied the startup and shutdown scripts for puppy2 in
detail, and made a suggestion how to better mount the pup_xxx.sfs file. For
certain PUPMODEs, if sufficient RAM, pup_xxx.sfs was copied from the boot
partition to pup_rw (the tmpfs ramdisk) then mounted by a loopback device
onto pup_ro2. Thus, pup_rw, the top unionfs layer, has a file that is mounted
as a lower layer, probably not the most healthy configuration. I cleaned this
up by copying pup_xxx.sfs to another tmpfs ramdisk, /mnt/tmpfs (path from the
point of view of the init script, prior to pivot_root). Thus pup_xxx.sfs is
in RAM but outside the unionfs layers.
Thanks to Lior Tubi (lior2b in Forum) who showed how to make the "Print
Screen" key work in the JWM window manager. I've added the appropriate line
into ~/.jwm/jwmrc-personal. Forum thread: http://www.murga.org/~puppy/viewtopic.php?t=6882
In the case of the multisession CD/DVD, the persistent storage is on the CD/DVD itself, so any extra .sfs files are left as-is and will be loaded at bootup ...but that is going to be slow when running Puppy. Can't help that though, as the whole idea of multisession is to be independent of the hard drive. I might add a bit more code to try and load them into RAM, if there's room.
In a nutshell, what the above paragraphs mean is that you can burn a puppy2 live-CD/DVD with the base iso file and any extra .sfs files that you want, and they will be automatically recognised and loaded.
I have setup a forum especially for developers and testers of Puppy2. This
is intended to be low-traffic and very focussed. Puppy2 is our experimental
next-generation Puppy, not yet for general usage. URL:
I had hoped to announce JWM upgraded to version 1.5, but the
taskbar-related bugs have got worse, not better. Here is a snapshot:
<!-- Additional tray attributes: autohide, width, border, layer, layout -->JWM 1.5 has lior2b's patch, and the horizontal space allocated for blinky should reduce to a very small space automatically? If I go back to width="39" for blinky, then behaviour is as the unpatched 1.4.
Also the "--disable-fribidi" configure option does not work.
Anyway, I'll report these problems to Joe, the JWM developer. If anyone else (Lior?) has any thoughts on this, let me know.
Um, I realised that the wording of the copyright notice at the bottom of
this page needs to be changed. It had something like "This is the logo for PuppyOS". Now it states "This
is a logo for Puppy, also known as
Puppy Linux and PuppyOS".
I've been thinking about the ownership of the various logos designed for
Puppy. Each Puppy site, and Puppy-derivative sites, are using their own
distinctive logos. It has evolved that way. It is possible to, that each
version of Puppy will have a logo with some distinctive feature that differs
from other releases. Anyway, would we be happy if someone took one of the
logos and used it for something else, perhaps totally unrelated to Puppy
Linux, or, another unrelated Linux distro?
I thought that I should do something, rather than nothing, but perhaps you
have thoughts about a better way to claim ownership of our logos? Please
raise the topic for discussion on the Puppy Forum. A good starting point is
to read about earlier input regarding Puppy logos: http://www.murga.org/%7Epuppy/viewtopic.php?t=3543
I would like to acknowledge David May, who vectorised a photo of the
"real" Puppy (see puppy_experiment2.png in the above thread). My logo, that
you see at the very top of this page, is my own take on converting the photos, inspired by David's work.
Foxti has progressed with getting VPN (Virtual Private Networking) with
MPPE to work. See this thread:
Apologies to anyone who sent an email to bkaulerATgoosee.com and it
bounced. This was due to the changeover of hosts. Or, it may have simply
disappeared -- I sent a couple of test emails and they were sent okay but
then just vanished. Anyway, the transition has completed and the email
address is working again.
Thanks to Jonathan Marsden (forum name jmarsden), we now have the Mantis bug tracking
Jonathan's Mantis bug tracker is here: http://tracker.nfshost.com/mantis
Note, the bug tracker is currently only for the "official" Puppy, not for
derived distros such as Grafpup. However, Puppy-based distros do sync with
the official Puppy more-or-less, but you would need to confirm a bug is also
in the latest official Puppy release before reporting it.
Domains puppylinux.com and goosee.com now moved to new host. Waiting for
the updated DNS records to filter throughout the Internet.
I received a renewal notice from my goosee.com host. As the account
expires in less than a month, and as Servage.net, the host for puppylinux.org
and puppylinux.com, is near CPU-usage capacity, I decided to signup with
another host and move goosee.com and puppylinux.com to it.
PupGet package manager is fixed to work in puppy2. There was a report on
the forum that it didn't uninstall properly, due to the absence of directory
/root/.usr -- that was easy to fix, but I had to rethink how files are
uninstalled if they are over-writing an original pristine file from
pup_xxx.sfs (equiv. to usr_cram.sfs). The way it was done in puppy1 could
actually corrupt the unionfs file management, but now it works fine. What
happens now, is if you install a package that has, say, library file xyz.so,
but puppy already has xyz.so in the pristine pup_xxx.sfs file, when the
package is uninstalled the original pristine file is restored. The exception
is a full hard drive installation -- you uninstall the package with xyz.so
it's gone, deleted. Well, that's how distros normally work anyway -- puppy's
restoration of the original pristine file is a special feature of Puppy.
The first dialog window that the Pupget package manager displays did not
look very professional. It used the xmessage program. I have reworked it to
One more thing. I remembered someone reporting that when the PupGet
package manager uninstalls a package, any empty directories (after the files
of the package have been deleted) are not deleted. Now they are.
A clarification about VPN from rarsa: "IPsec VPN is
happily running in puppy. I have been using it for quite some time. It's just
the MPPE VPN people have had trouble with." More details on Forum:
Well, it looks like I will finally be getting
broadband. Here in Australia the government has a subsidy scheme to bring
broadband to rural areas, and my local town is getting wireless broadband,
via satellite. I'm not yet 100% sure that my place will be able to connect
up, as I'm out of town a bit and not quite line-of-sight. Anyway, fingers
crossed. It will solve my uploading problems. I'm not sure when they will be
installing, maybe one or two months away.
I'm in the process of setting up a blog. I have
sometimes referred to this Developer News page as my blog, but it's really
only a plain static HTML page. The blog I'm using, WordPress, looks real good
from the description. It's a bit like a wiki in that others will also be
allowed to contribute. Anyway, I don't really know much yet as I've only got
as far as installing it. I know my web pages are badly in need of an
overhaul, so I'll play with WordPress and see if it is what I need. If anyone
reading this has experience with WordPress and would like to comment, or
maybe give me a suggestion about some feature of WordPress that would be good
for my site, please post on the Forum and we'll have a chat.
The 2.4.31 kernel has been recompiled with the mppe-mppc patch version
1.3. So, the kernel now has three patches, ntfs, squashfs and mppe-mppc.
ppp has been upgraded to version
2.4.3 and patched with mppe-mppc patch version 1.1.
GUI frontend for wget downloader is now in Puppy. Version 0.4 has a bug, so I
have named the bugfixed version 0.4.1 and made it into a Unleashed
I've been discussing how to implement VPN in Puppy, with Foxti on the
Forum. The kernel needs to be patched to provide MPPE support -- note, kernel
2.6.15 has MPPE support, but I'm using the 2.4.31 kernel for puppy2 (and
maybe for v1.0.9). Looking for patches, I found this site: http://mppe-mppc.alphacron.de/.
This seems to be what we need, so today I plan to recompile the kernel, and
patch and upgrade ppp.
Note, tempestuous has compiled MPPE kernel modules for the 2.4.29 kernel
currently used in Puppy. See Forum thread: http://www.murga.org/~puppy/viewtopic.php?t=6606
Thanks to jmarsden who pointed out how to enable the mouse for Xlock.
Previously, a keypress was required to unlock the screen. Now, a mouse motion
also will do it. (The password of course still has to be entered to complete
Thanks to some great detective work, the bug in Xlock is now fixed. Xlock
is a desktop icon on the top-right of the screen -- click on it to lock the
screen and display a pretty pattern (right-click to choose a pattern). Unlock
requires a password, which was the problem, as any password unlocked the
Puppy's little CD/DVD burner program, burniso2cd, could not handle USB CD
drives. That is now fixed. In fact it should now also handle SCSI CD/DVD
I'm fiddling around with the usb keyboard and mouse setup. There were
reports of trouble on the Forum. One thing that I have done is migrated from
the usbkbd.o and usbmouse.o kernel modules to the hid.o, mousedev.o and
keybdev.o modules. The first two use a simpler system known as HIDBP (Human
Interface Device Boot Protocol), whereas the latter use a more sophisticated
HID protocol. I'm using a usb keyboard and mouse right now, seems okay.
Also right now I'm running from a multisession-CD -- that seems to be very
solid now. Will keep testing it, make sure it stays that way.
Upgraded rarsa's improved Network Wizard script to version 21.
Note, having these drivers builtin, rather than having to download and install them, is good for dialup people, however of course these are not the only kind of software modems. If you have a HSF/HCF type of modem, you still have to install the driver separately -- see the Forum.
(As always though, we recommend you get a real hardware modem -- second-hand external serial modems are very cheap on eBay).
I rewrote the script for the desktop "Save" icon for puppy2. For
multisession-Puppy, the "Save" icon is a means to save the session to CD-R/RW
or DVD-R immediately, without having to wait until shutdown.
Why would you want a "Save" button? Previous Puppy users found that saving
at shutdown is iffy, so this gets around that problem -- although the latest
puppy2 saves very reliably at shutdown. It can also be useful if you are
writing the next Great American Novel and worry about a power outage causing
loss of the last three hours of typing.
Note, previous puppies didn't have a "Save" icon for multisession-CD,
however do have it now, but it needs to be used sparingly as CDs fill up very
fast. This is due to the considerably less storage on a CD, but also each
saved-session has about 12M (I think, from memory) wasted space (whereas a
saved-session to DVD is much more space-efficient).
Technical note: the actual script is /usr/sbin/savesession-dvd.
Major improvements to multisession shutdown. Reliability of session-saving
for both CD and DVD improved.
Note that all notes in this page from now on refer to puppy2, unless
specifically stated otherwise.
Good news. I found that the problem with ndiswrapper and the 2.4.31 kernel
is the lck patchset (from plumlocosoft). I took the vanilla 2.4.31 source and
only applied the ntfs 2.1.6b and squashfs 2.2 patches, and now ndiswrapper
works. So, we are back onto using the 2.4.31 kernel for puppy2.
Ndiswrapper is upgraded to version 1.10.
Pakt (forum name) has made further improvements to xkb, the keyboard layout component of Xorg. If you find that the layout is still wrong for your country, visit the Forum here:
also here, for a different solution from MU:
For developers, puppy-unleashed-core-1.0.8r1.tar.gz has been uploaded to ibiblio.org.
version 1.0.8r1 (revision 1) released. This is a few minor bugfixes.
Each of these bugs in 1.0.8 can be manually fixed (see notes below), however
for convenience here is a release with the fixes. The size is 61.7M. Release
Peter Sieg has also mirrored Puppy 1.0.8:
Another one snuck through, also easily fixed:
Please download the file xkbcomp.gz, to /usr/X11R6/bin/. You can uncompress it by opening a terminal (Running Rox file manager, with /usr/X11R6/bin as current window, right-click and choose "xterm here"):
# gunzip xkbcomp.gzThen restart X (look in Shutdown submenu).
If you find this still does not work for your country, there is another solution, that also works for Puppy 1.0.7:
Puppy 1.0.8 is mirrored here:
Ted Dog has mirrored them here:
We will bring out a "service pack" to fix any minor bugs, but this one is easy to fix manually.
When you look at the menu in 1.0.8, near the bottom is the "Help" entry. This is not working.
Please start the Rox file manager, right-click the mouse to bring up a menu, choose "Display -> Show hidden". Then you will be able to see hidden files. Click on the file ".jwmrc". There is this line:
<Program label="Help" icon="help24.xpm">LD_PRELOAD=/usr/lib/libgdkxft.so dillo /usr/share/doc/index.html</Program>Change it to this:
<Program label="Help" icon="help24.xpm"> dillo /usr/share/doc/index.html</Program>Save the file, then choose "Shutdown -> JWM restart" from the menu. That should fix it.
CLARIFICATION: Using Gaim
If you want to start Gaim and be automatically connected to the #puppylinux IRC chat channel, please start Gaim from the menu the first time that you run Gaim. The desktop icon just starts Gaim without the automatic account creation -- but after having run Gaim once from the menu it will then be ok to run from the desktop icon as the account has been created. Slight inconsistency there!
|(c) Copyright Barry Kauler 2008. All rights reserved. http://puppylinux.com