Saluki

[scsijon]

If we call Saluki Puppy the start of the 6.x Series, as I think we should consider it to be with it's proposed ideas and function leads starting to appear, we can consider you getting into the fray with 6.1 with your link overlay method included!

I think calling it Puppy 6 is probably excessively ambitious and maybe a bit presumptuous.

jemimah, you and tman are moving into new structure steps with puppy, and those steps are when Barry has gone up a number, maybe I should have said 5.99, however the Saluki thread plus this one makes me think of how he worked when stepping from 3 to 4 and 4 to 5.

I do believe, from what you have achieved in the past and taking on the lead of Saluki so soon after a return, that you have the ability to lead into the next step! However, it WILL always be up to Barry whether he decides it's far enough to actually be called Puppy!

And that's enough on this topic from me!

And thank you Technosaurus, ?maybe that should go into the Developers Thread also and see what happens. Another step forward for puppy I believe!

regards
scsijon

[Sage]

And that's enough on this topic from me!

& co.
Angel has the skills, enthusiasm and energy to become the new BK when he finally hangs up his boots and sets off with his gold-panning dish into the Outback. Preferably this would occur gradually with massive prior cooperation between the two of them and an ongoing permanent personal satellite phone link between them? Probably a widely approved succession. One has already made an outstanding contribution, the heir apparent will continue to do so.

[technosaurus]

And thank you Technosaurus, ?maybe that should go into the Developers Thread also and see what happens. Another step forward for puppy I believe!

regards
scsijon

Right now I am just exploring various tangents and I have been in the "write one to throw away" phase.

for instance, my latest variation on multicall binaries:

I got to thinking about adding many apps to the mcbs and the multiple "if ... strcmp ..." will start to take some time and add size

I wanted to do a switch case, but you can only do 1 character with it or an integer etc... so I found a way to convert the strings

change this:

Code: Select all

echo 'int  main ( int argc, char** argv ){
#define _(x1,x2,x3,x4)  (((((((x4)<<8)|(x3))<<8)|(x2))<<8)|(x1))
switch ( _(argv[0][0],argv[0][1],argv[0][2],argv[0][3]) ) {' >main.c

and for the applets, just use the first 4 letters as such:

Code: Select all

echo 'case _('j','w','m','\0')	:  return(jwm_main(argc, argv));' >>main.c

since _() is a macro all of the applet names get defined at compile time as a single value instead of an array of characters the only thing that gets checked at runtime is some simple math on arg0 and 1 comparison instead of X*#of applets

Did I miss anything important?
I know busybox must do something similar because there was a recent post for fixing swapon and swapoff, so need to mention that in comments I guess (its an easy fix with another switch case if we ever get names that close)

There are a few apps that share some large and uncommon libraries that could benefit from a partial static build:
ffmpeg+ffplay/*mplayer (libav*)
parted+gparted+e?fsprogs (libparted, libuuid etc...)
ttuuxxx could probably tell us a bunch of others

@Lobster - thought I saw you asking about this somewhere...
I came up with a better way to do flat a file database while exploring project management:
http://www.murga-linux.com/puppy/viewto ... 711#582711
the gist of it is
####################bzipped database file
VAR1="something" ... VAR99="another"
...
VAR1="something else" ... VAR99="another thing"
####################bzipped database file
Note: VARs can be named anything, not just VAR*

which can very quickly be manipulated like this:

Code: Select all

bzcat $FILE | while read LINE ...
eval $LINE
#code to manipulate entries here
echo 'VAR1="something" ... VAR99="another"'
done |bzip2 -9c >$FILE~ && mv -f $FILE~ $FILE

The other big one that needs code review from someone who actually cares is localization (as a native english speaker, I really don't, I just found other implementations such a laughable PITA, that I had to try pulling on the sword)
The auto-translate part could use some help from a multilingual coder:
http://www.murga-linux.com/puppy/viewto ... 998#583998
it is certainly a lot easier and faster than gettext, but its just not something I am passionate about. Fortunately L18L is championing this cause

I am still poking around on the jwm_tools from time to time, but I already have it doing everything but the tray (which is already well covered by other apps) so it should ready for a 1.0 release soon if it gets some testing:
http://www.murga-linux.com/puppy/viewtopic.php?t=70804

Anyhow, I am posting all of this because ... inspired by Rob Landley's BSD relicensing of toybox, I am going back to work on bashbox, rewriting it from scratch this time for compatibility with toysh and so I can also release it under a UIUC (BSD-ish) license ... if I get a wild hair I may even port thttpd to toybox so I can do more work on my cgi script code
... I will still be around for mentoring, but not a lot of maintaining or recompiling - so if anyone wants to take up being maintainer of any of the various packages I have compiled, I'd be more than willing to pass on the knowledge/tips/tricks that I use to make them smaller and compatible with multiple versions of Puppy (hint: autotools/pkg-config is screwing us)

oh yeah -which reminds me of another unfinished tangent - a gcc wrapper to keep autotools/pkg-config from screwing us:

Code: Select all

#!/bin/ash
#BEGIN CONFIG##################################################################
#edit these to taste :)
XTRINCS="-I/usr/src/include"
XTRCFLGS="-ffunction-sections -fdata-sections -mno-accumulate-outgoing-args -fvisibility=hidden"
XTRLFLGS="-L/usr/src/lib -Wl,--gc-sections,--as-needed,-s"
STATIC=-static
ARCH=""
TUNE=""
CC=/usr/bin/gcc
#see first case statement to blacklist CFLAGS
#END CONFIG####################################################################

for ARG in $@ ; do
[ ! "$RESET" ] && RESET=1 && set --
case "$ARG" in
	-g*)continue #add blacklisted CFLAGS here
	;;
	-O[0-3])set -- $@ -Os
	;;
	-fPIC|-fpic|-fpie|-fPIE|-DPIC)[ "$STATIC" ] && continue || set -- $@ "$ARG"
	;;
	-march=*|-mcpu=*)[ "$ARCH" ] && set -- $@ -march=$ARCH || set -- $@ "$ARG"
	;;
	-mtune=*)[ "$TUNE" ] && set -- $@ -mtune=$TUNE || set -- $@ "$ARG"
	;;
	-l*|-o|-Wl*)LINKING=true && set -- $@ "$ARG"
	;;
	*)set -- $@ "$ARG"
	;;
esac
done

[ "$LINKING" ] && LD_FLAGS=$XTRLFLGS

echo $CC $XTRINCS $XTRCFLGS $STATIC $LD_FLAGS $@ $LD_FLAGS
$CC $XTRINCS $XTRCFLGS $STATIC $LD_FLAGS $@ $LD_FLAGS

this is currently just a stub ... I planned to add more facilities for different toolchains like uclibc, musl libc, and another for uclibc++

[Lobster]

Angel

Must have fallen off the Christmas tree whilst no one was watching . . .

These are the known aims of Puppy Saluki, which can be found here:
http://puppylinux.org/wikka/Saluki

• Target systems will be modern to semi-modern hardware
  ISO size should be 150MB or smaller
  Based on Racy 5.2.2 or Wary with a new xorg and kernel
  Focus will be user friendliness, innovation and artistic look and feel
  Solid, functional applications
  IceWM and possibly PCManFM2 Window managers

If any of that changes let me know

Whilst we await the return of Tman and Luki 003, someone might like to try the new Libreoffice SFS in Saluki . . .
http://bkhome.org/blog/?viewDetailed=02619

Want a simple Saluki database which will work in all Puppys? - might need to dust off your coding wings . . .
http://puppylinux.info/topic/p-data-alp ... e-database

Thanks Jemimah, thanks guys

[Billtoo]

I did a manual frugal install to an SDHC card formatted fat32.
I installed the devx sfs and compiled the ati-driver-installer-11-11-x86.x86_64.run

It's working well so far.

[jemimah]

oh yeah -which reminds me of another unfinished tangent - a gcc wrapper to keep autotools/pkg-config from screwing us:

Code: Select all

#!/bin/ash
#BEGIN CONFIG##################################################################
#edit these to taste :)
XTRINCS="-I/usr/src/include"
XTRCFLGS="-ffunction-sections -fdata-sections -mno-accumulate-outgoing-args -fvisibility=hidden"
XTRLFLGS="-L/usr/src/lib -Wl,--gc-sections,--as-needed,-s"
STATIC=-static
ARCH=""
TUNE=""
CC=/usr/bin/gcc
#see first case statement to blacklist CFLAGS
#END CONFIG####################################################################

for ARG in $@ ; do
[ ! "$RESET" ] && RESET=1 && set --
case "$ARG" in
	-g*)continue #add blacklisted CFLAGS here
	;;
	-O[0-3])set -- $@ -Os
	;;
	-fPIC|-fpic|-fpie|-fPIE|-DPIC)[ "$STATIC" ] && continue || set -- $@ "$ARG"
	;;
	-march=*|-mcpu=*)[ "$ARCH" ] && set -- $@ -march=$ARCH || set -- $@ "$ARG"
	;;
	-mtune=*)[ "$TUNE" ] && set -- $@ -mtune=$TUNE || set -- $@ "$ARG"
	;;
	-l*|-o|-Wl*)LINKING=true && set -- $@ "$ARG"
	;;
	*)set -- $@ "$ARG"
	;;
esac
done

[ "$LINKING" ] && LD_FLAGS=$XTRLFLGS

echo $CC $XTRINCS $XTRCFLGS $STATIC $LD_FLAGS $@ $LD_FLAGS
$CC $XTRINCS $XTRCFLGS $STATIC $LD_FLAGS $@ $LD_FLAGS

this is currently just a stub ... I planned to add more facilities for different toolchains like uclibc, musl libc, and another for uclibc++

This is great. Thanks!

[James C]

Saluki 002 live pfix=ram.Sound and internet working,had to use xorgwizard to get desired resolution (1440x900).All the basics looking good.

# report-video
Saluki, version 0.0.2 on Sat 10 Dec 2011

Chip description:
0.0 VGA compatible controller
nVidia Corporation NV18 [GeForce4 MX 440 AGP 8x] (rev c1)
oem: NVidia
product: NV18 () Board Chip Rev A2

X Server: Xorg
Driver used: nouveau

X.Org version: 1.11.0
dimensions: 1440x900 pixels (380x238 millimeters)
depth of root window: 24 planes

...the above also recorded in /tmp/root/ as report-video,
and archived with xorg.conf and Xorg.0.log as report-video-full.gz


# free
total used free shared buffers
Mem: 1034016 468272 565744 0 63548
-/+ buffers: 404724 629292
Swap: 1228936 0 1228936
#

[mikeslr]

Hi Tman & jemimah:

Welcome back jemimah. You've been missed.

To PAE or not to PAE: Racy or Quirky?

A couple of posts back Lobster quoted the stated objectives of Saluki, including the then open question whether it was to be based on Quirky or Racy.

I do most of my exploring-experimenting on a Quad-Core, 4 Gb outfit. If ever I stumble upon some combination I think might be an improvement over a system running on one of my older computers, I port it over. I hadn't been following the Quirky/Racy threads. Thinking that Racy 5.2.2 might provide some advantage over the dpup/upup/spup's they're presently running, I started with Racy and worked up a remaster on my Quad-Core. Only to discover that it wouldn't boot on my second most often used computer, a relatively recent Thinkpad T42. No PAE.
What's this PAE? I asked. And, how do you turn it off? Wikipedia provided an answer to the first question, and goggling revealed, you can't. If a system's kernel is built PAE enabled, you're stuck with it. You're options are not to use that OS, or replace the kernel: the latter being neither an intuitive nor nooby-friendly process.
So I asked the next obvious question: how significant a performance advantage does PAE provide? Perhaps my search skills failed me. I could only find two posts giving a comparison of test results of PAE-enabled and not. Both were run using Ubuntus (Ubunti?).
http://www.phoronix.com/scan.php?page=a ... _pae&num=1
http://www.phoronix.com/scan.php?page=a ... ae64&num=1
Despite those reports, Ubuntu is considering dropping support of non-PAE kernels. Maybe Ubuntu knows something I don't. Ubuntu, however, is also pushing the Unity Desktop. So I have to wonder about the state of Ubuntu's mental health. But then I am, by nature, a “Radical-Middle-of-the-Roader.

[Tman]

Hi Tman & jemimah:
To PAE or not to PAE: Racy or Quirky?

mikesir
Jemimah and I have already agreed that a non-PAE build is the way to go. The next step would be to compile the new kernel. Jemimah could do this, but it would save us time if we could get someone else to build a kernel for us.

[James C]

Just playing around and put together a Saluki-0.0.2-k-3.10.Running live, pfix=ram it boots much faster.


Kernel : Linux 3.1.0 (i686)
Compiled : #1 SMP Sat Nov 12 21:58:58 EET 2011
C Library : GNU C Library version 2.10.1 (stable)
Default C Compiler : Unknown
Distribution : Unknown distribution
-Current Session-
Computer Name : puppypc6392
User Name : root (root)
Home Directory : /root

# report-video
Saluki, version 0.0.2 on Sat 10 Dec 2011

Chip description:
0.0 VGA compatible controller
nVidia Corporation NV18 [GeForce4 MX 440 AGP 8x] (rev c1)
oem: NVidia
product: NV18 () Board Chip Rev A2

X Server: Xorg
Driver used: nouveau

X.Org version: 1.11.0
dimensions: 1440x900 pixels (380x238 millimeters)
depth of root window: 24 planes

...the above also recorded in /tmp/root/ as report-video,
and archived with xorg.conf and Xorg.0.log as report-video-full.gz
#

[Tman]

James,

Where did you get the 3.10 kernel from? Is it Pemasu's compile?

[James C]

James,

Where did you get the 3.10 kernel from? Is it Pemasu's compile?

Yes, I used Pemasu's kernel compile.It works great and it's well tested. No sense reinventing the wheel.

[l2ulinux]

I downloaded Saluki 002 and burn to CD. Booted with no problem, ran with no problem and everything went well. It ran great.
Misc
Uptime 11 hours, 16 minutes
Load Average 0.01, 0.02, 0.05

I saved to CD because I always save and then reboot to see if everything saved OK.
Input Devices:
Microsoft Comfort Curve Keyboard 2000
Logitech USB-PS/2 Optical Mouse
Mouse works great but keyboard will not work.

I rebooted without running the save file and it works great. I am running it now to write this.

[pacer106]

I toned down the brightness.


saluki9-1366x768 by J. P. Stunna, on Flickr


saluki9-1024x768 by J. P. Stunna, on Flickr

[MinHundHettePerro]

I toned down the brightness

Stunning! (In actual use, the previous ones would've hurt my eyes, I reckon.)
Cheers /MHHP

[nancy reagan]

Hi Pacer,

Wouldn't it be more practical to start a separate topic for the Saluki eyecandy ?

Then you have them at one glance and besides it does not take so much room of the more technical details.

Like your wallpapers, though -at least my opinion - prefer a more smooth background without the waferpattern. But then again that is my personal experience.

Keep up the good work.

[jim3630]

Whilst we await the return of Tman and Luki 003, someone might like to try the new Libreoffice SFS in Saluki . . .
http://bkhome.org/blog/?viewDetailed=02619

crashed couple of times shutting down to X but able to restart puppy with "xwin." opening the document chose yes to save crash report but can not find one. did get a mention or error in xerrs.log. ?lack of java. remove phony gz to see report.

[rjbrewer]

Hi Tman & jemimah:
To PAE or not to PAE: Racy or Quirky?

mikesir
Jemimah and I have already agreed that a non-PAE build is the way to go. The next step would be to compile the new kernel. Jemimah could do this, but it would save us time if we could get someone else to build a kernel for us.

An interesting thread on switching kernels that may be of
use to us that can't use a PAE kernel.

http://www.murga-linux.com/puppy/viewtopic.php?t=60180

[Billtoo]

Jemimah and I have already agreed that a non-PAE build is the way to go. The next step would be to compile the new kernel. Jemimah could do this, but it would save us time if we could get someone else to build a kernel for us.

The luki-0.0.2 release is a keeper.
I look forward to trying out any future releases no matter what
kernel is used.
Thanks

[jemimah]

James,

Where did you get the 3.10 kernel from? Is it Pemasu's compile?

Yes, I used Pemasu's kernel compile.It works great and it's well tested. No sense reinventing the wheel.

I think Pemasu has a non-PAE kernel somewhere. I'd prefer to collaborate on this since if we can agree on a kernel, we can share drivers. It's less work for everyone.

Try the one here: http://www.murga-linux.com/puppy/viewtopic.php?t=71755