This build has the simplified filenames, plus a separate 'zdrv' SFS file.
Download from here:
I would like to get feedback on this kernel, what do you think about it. Does it fix any problem, that you had with an earlier Wary, for example sound?
I have what may seem like a strange request. After booting, the boot media has 'zdrv.sfs' -- can you open that directory in a terminal window and execute 'md5sum zdrv.sfs' and confirm that the md5sum is 'a3a7c54901fc09b296d284af625041fa'.
The reason for the above odd request is that I have a PC that gives the wrong md5sum sometimes, but only with the 18.104.22.168 and the 22.214.171.124+ kernels. I have booted up with other kernels and got the correct md5sum. I have not been able to determine the cause of this, but I am hoping that it is due to a hardware fault in the PC, and noone else will encounter it.
At this stage, I think a peculiar hardware fault is the most likely explanation, though why it should show up only with certain kernels I don't know. The 'tar' operation also misbehaves -- perhaps there is a dicky RAM memory bit that certain kernels are not using (but, I have used a memtest boot CD to test the RAM and it was ok).
The final Wary
I am thinking for the final release, probably version 5.1, I will do it like the "mix and match" that I did for Wary 070 (see http://distro.ibiblio.org/pub/linux/distributions/quirky/test/wary-070-mix-n-match/).
But, also have one ready-built live-CD ISO file with say the normal recommended kernel (126.96.36.199?).
Then there is the question whether I should go for the simplified filenames (vmlinuz, initrd.gz, puppy.sfs, zdrv.sfs) in the final.
Comments:Posted on 9 Feb 2011, 8:08 by technosaurus
I like the idea of separate kernel options in the zdrv because I use the zdrv cutter to cull the cruft. There are a couple of versions of the zdrv cutter that amigo helped me put together floating around these days. It cuts the zdrv down to ~2-3Mb. If you are interested I can clean it up and patch the installer to _optionally_ use it (with a big warning that they need to have all removable devices plugged in and working that they might use later) ... probably best suited for quirky - maybe even modify it to go directly to the save file?
filenames short/long don't really matter too much as long as the psubdir etc... is settable (the unique identifiers tacked on to the end on the other hand seemed to make things a lot more complicated)
Posted on 9 Feb 2011, 8:07 by BarryK
Re zdrv name
In recent Woof-built puppies, the filenames are in /etc/DISTRO_SPECS:
#16-byte alpha-numeric ID-string appended to vmlinuz, puppy.sfs, zdrv.sfs and devx.sfs:
#Puppy default filenames...
#Note, the 'SFS' files below are what the 'init' script in initrd.gz searches for,
#for the partition, path and actual files loaded, see PUPSFS and ZDRV in /etc/rc.d/PUPSTATE
...so it doesn't matter what the names are, scripts can lookup DISTRO_SPECS to see what they are. Also can look in /etc/rc.d/PUPSTATE.
Posted on 9 Feb 2011, 10:57 by james c
The md5sum of the zdrv.sfs matches here.
# md5sum zdrv.sfs
Posted on 9 Feb 2011, 11:41 by 01micko
matches here too
Posted on 9 Feb 2011, 11:48 by broomdodger
md5sum matches here also, acer travelmate 4670 notebook
Processor : 2x Genuine Intel(R) CPU T2300 @ 1.66GHz
Memory : 1025MB (173MB used)
Operating System : Puppy Linux 0.50
User Name : root (root)
Date/Time : Tue 08 Feb 2011 07:42:26 PM PST
Posted on 9 Feb 2011, 12:46 by broomdodger
md5sum matches here also, dell inspiron 4100
Processor : Intel(R) Pentium(R) III Mobile CPU 866MHz
Memory : 255MB (61MB used)
Operating System : Puppy Linux 0.50
User Name : root (root)
Date/Time : Tue 08 Feb 2011 08:40:53 PM PST
Posted on 9 Feb 2011, 14:07 by Raffy
Yes, Barry, simplified filenames will be good especially for user-oriented documentation.
Posted on 9 Feb 2011, 14:21 by scsijon
1/ just tested 502, problem listed there,will download 503 tomorrow and test.
2/ Since your talking about 5.1. I wonder if you could do a mixnmatch type package inside the iso, with a default (minimal) build to start it (with warnings), do a modprobe as part of the install routine, the results of the modprobe giving a list of suitable kernals and user asked which one to install with one of them marked as "most suitable for your hardware combination".
and no, please for simplified filenames, it makes it too hard to work out what belongs with what, especially when having more than one version frugally installed on a single partition.
Posted on 9 Feb 2011, 15:40 by BarryK
Re simplified filenames
I know that having 'puppy.sfs' instead of 'wary_503.sfs' and 'zdrv.sfs' instead of 'z5033228.sfs' might seem to be a potential problem with getting them mixed up when different installations are in the same partition, however we already have 'vmlinuz' and 'initrd.gz' that have no version information.
So, in practice, having the names 'puppy.sfs' and 'zdrv.sfs' should not make things any more difficult or confusing.
But we do need some experience with this to verify.
Posted on 9 Feb 2011, 15:53 by Terryphi
Locale settings still lost on upgrade
md5sum for zdrv.sfs as expected.
Saved locale settings are still ignored on upgrade of warysave.2fs.
Posted on 9 Feb 2011, 16:23 by BarryK
Re locale lost
Are we talking about the same thing... I thought that I had fixed the LANG variable in /etc/profile so that it does not get overridden at an upgrade.
Has the LANG= line in /etc/profile definitely been set back to en_US for you?
Posted on 9 Feb 2011, 16:32 by Terryphi
On upgrade the choose locale dialog appeared offering en_US. This suggests to me that the script did not work or did not like what it found in etc/profile.
Is it relevant that I use en_GB.utf8? Does your script change in /etc/rc.d/rc.update make provision for utf8?
Posted on 9 Feb 2011, 16:34 by BarryK
Wary 503 forum feedback
The forum feedback thread is here:
...503 reports start on page 2 of the thread.
Posted on 9 Feb 2011, 18:27 by Terryphi
I have now checked upgrading en_GB (not utf8) and the LANG= line in /etc/profile is set back to en_US.
Posted on 9 Feb 2011, 18:34 by Sage
Put together some weird HW combos - they all give the correct checksum.
Posted on 9 Feb 2011, 18:37 by Sage
Terry: is that a FULL or FRUGAL you are using?
Posted on 9 Feb 2011, 18:44 by Terryphi
Sage: I'm always Frugal. ;)
Posted on 9 Feb 2011, 18:47 by Sage
Would this be a good moment to include Midori, Opera11 or Chrome in the list? And to rectify the strange misplacementof the Opera logo in the menu list, previously reported many times?
Posted on 9 Feb 2011, 18:58 by Sage
...never appealed to me - all those .sfs' to confuse the issue with their ever-changing nomenclature and enumeration!
Same issue with folks who want to duel(sic) boot with 'doze (can you believe it?!) and run Wine. All we get on the Forum is their whining! Time to liberate a few dozen old HDs from landfill, buy a few caddies, run one HD per distro. OK - keep one for Windydoze, too, if you really must.
Notwithstanding, it's great fun to find out the consequences of mix'n-matching OSes that were never intended to meet (specifically intended to avoid such in the case of 'dozey) - I do it all the time for a laugh - but not, as some of the folks on the Forum imply, to run it on mission-critical (sole?!) box. That route is sheer folly and developers would save themselves a lot of tears by spelling it out. Exceptionally, Jemimah Jewel seems to enjoy rising to their every demand! Such is youth....
Posted on 9 Feb 2011, 19:38 by FeodorF
could you please give any hint about your PC with the wrong md5sum (CPU, ram-size, chip set). Thank you.
Posted on 9 Feb 2011, 20:52 by ICPUG
The argument put forward for simplified filenames, that initrd.gz and vmlinuz have always been simplified so why not puppy.sfs, seems irrefutable as a reason but not as giving any benefit.
I personally favour some version tag on the filename because there is the benefit that you know what puppy distro it comes from without jumping through hoops executing some command to find out. What benefit is there for simplified filenames that we can trade off against?
If the similarity of all filenames was the only benefit I would say add the version to the filenames of initrd.gz and vmlinuz rather than remove it from puppy.sfs! (This is said with tongue in cheek - just in case it is misconstrued as whining)
Posted on 9 Feb 2011, 22:46 by carolus
Found this somewhere, in connection with the wireless chipset in my newest computer:
"Any distribution shipping a kernel >= 2.6.27 will have ath9k present but the ath9k driver on 2.6.32 is the oldest one recommended, anything older than that is completely unsupported"
Posted on 10 Feb 2011, 3:33 by rodin.s
I think mix-and-match is a good idea. I needed to choose kernels for my old PC because of ISA-soundcard and long boot and also for my home PC where 2.6.33 of Lupu (or Quirky) hangs up (have to reboot with reset button) when I hotplug in or out 3g-modem or USB-flash. So if simplified filenames is needed for kernel switch than it's OK.
Posted on 10 Feb 2011, 8:11 by Raffy
simplified file names
Newbies only try one Puppy at a time, so a simplified naming of files will help. Advanced users should be able to name/identify the psubdir where they put the files.
But just in case, perhaps someone can hack a script that reveals the version number of Puppy files in a given folder?
Posted on 10 Feb 2011, 12:14 by scsijon
this n that
found cause or problem with 502 (and 503) above, see forum page 3
also ps comments on zdrv and pc may be of help
would still like full filenames, I had to hash the puppy.sfs's to work out what had actually happened rather than just look at them.
Note for others: I may add that initrd.gz and vmlinuz are linux independant, the names don't belong to any particular variety, there the linux standard. I can use my suse rescue cd to boot puppy by telling it where to look for them and believe the other varieties work the same.
Posted on 10 Feb 2011, 15:47 by Mark Forrest
this is the first puppy apart from the updated 2.xx and Jemimah's fluppy that have boooted on my Aspire One 751h. Usually they stick at "loading drivers to access drives" I belive this is because it doesn't have a true Sata interface, but a Sata to Pata adapter!
I'm having to use Xvesa due to the lack of Pouslbo drivers and sound isn't working yet, but at least I can play now.
Posted on 12 Feb 2011, 3:59 by 785ljk
md5sum, sound, kernels - HP Pavillion 4473
Computer: HP Pavillion Model 4473.
Processor: Celeron 400 MHz.
Memory: 256 MB.
Sound: Rockwell Int. Riptide (on motherboard).
"zdrv.sfs" md5sum was correct as listed above.
Possibly the following may help as far as sound goes for people who know much more than I do. On the above Pavillion machine using the following kernels with ALSA (various Puppies and other distros), the sound "card" seemed to be listed correctly in the PCI hardware; but ALSA could NOT recognize the card:
Again, with the same above Pavillion machine, sound WORKED with distros using the kernels below. ALSA recognized the sound "card":
188.8.131.52 Tiny Core Linux, ver. 2
184.108.40.206 Puppy 4.3.1
220.127.116.11 Slitaz 3.0
18.104.22.168 Wary Puppy 5.03b
Thank you all for all your work. Take care.
Posted on 13 Feb 2011, 5:04 by rodin.s
switch keyboard in urxvt
It's known issue. urxvt needs LANG=ru_UA.UTF-8 to switch keyboard layouts. With LANG=ru_UA.utf8 (default) - only english layout works. And always some hacks needed to make it work.
Posted on 16 Feb 2011, 8:05 by BarryK
Re locale lost
Ok, I think that I have fixed it, for the next Wary release.