test results for alpha puppy2

[BarryK]

In p2 (Puppy2) I notice that @ has to be typed in using "

Puppy is set up for US so that is not right?
I did run the locale thing - so please test if on a pristine p2

Lobster, you have a British keyboard? Where are the " and @ located on your keyboard?

Pup2alpha is using Xkb, but it needs to be setup manually.
If you look at the Developer News page, somewhere about Feb 12 I made a note on this. Your layout would be "gb", yes?
Looking in /etc/X11/xkb/symbols/, there is a "gb" in there, so add lines into the keyboard section of /etc/X11/xorg.conf like this:

Code: Select all

   Option "XkbModel" "pc102"
   Option "XkbLayout" "gb"

...at least I think that would be right.

[BarryK]

I thought I'd give Puppy2-alpha a try. It was not successful.

First boot. Everything went as expected until Puppy2 tried to find my sound card. First I tried to identify the card that I had. It did not find it, and went on to select video mode. I chose Vesa, and Puppy2 decided to take another look for my sound card. This time I just enterd my way through and Puppy2 got hung up when Alsa tried to identify my card.

I finally had to reboot.

Second boot, the same as the first, except this time I told Puppy2 not to search for any legacy sound cards. Again it wanted to search for a sound card after I chose the VESA mode, but finally booted the screen.

Since I normally do a HD2 install that was my next step. Puppy2 went throught the process of installing itself. Except for GRUB. I accepted the defaults, including the one where it asked where to put Grub. The default was "NONE" It told me it was working but nothing was happening.

When I went to reboot, Puppy2 asked if I wanted to save a file, to disk. I chose HDA!, vfat, but nothing was saved.

Finally re-installed 1.0.6

Yes, we have identified that bug with alsa. It is okay if the correct card is found. But if not, alsaconf will run everytime X is started.

When you get to that part of the h.d. install where Grub is installed,
accept all defaults, except for when it asks where to put Grub bootloader -- choose MBR.

[BarryK]

GuestToo,
I hadn't got around to testing a porr-man's h.d. install.
Will do so promptly.

What about booting from live-CD on your PC, is that working?

[tempestuous]

I had hoped to avoid using a boot menu.

Barry, I suggest you could still avoid a boot menu, but have a prompt with a 5 second delay for the user to add any boot options if needed.
Most Linux installers I have used work this way. I think (?) the isolinux.cfg would be like this -

default puppy
display boot.msg
prompt 1
label puppy
kernel vmlinuz
append root=/dev/ram0 initrd=image.gz
timeout 50

The boot.msg would say something like -
"If you need particular boot options like "acpi=off" type "puppy" then a space, then each boot option separated by a space."

"puppy" is obviously the boot-choice-label. You must type this text first, otherwise isolinux will think that the first boot option you type is a boot-choice-label, and will report this as an invalid choice. Most Linux distros I have seen label their default boot choice "linux" ... so you could use this label instead of "puppy" in the interests of standardization.

Just recently I booted Puppy 1.0.7 CD and added a boot option in this manner (vga=788) ... it worked fine.

I think that "acpi=off" would be OK with ACPI-enabled computers (and can sometimes make previously unrecognised hardware work) but I think that some users would then complain that certain features don't work - like power saving features, battery meters etc.

[Trobin]

Yes, we have identified that bug with alsa. It is okay if the correct card is found. But if not, alsaconf will run everytime X is started.

When you get to that part of the h.d. install where Grub is installed,
accept all defaults, except for when it asks where to put Grub bootloader -- choose MBR.

Thanks. But the first two tries didn't get that far. I did have to change one default setting in Grub Install. Where it asks where to put the Grub Files, the default is "NONE." I changed it to /dev/hda2, and it worked. Xorg did not. Which is okay as Xvesa did.

[dewdrop]

Hi Pups,

Here's my 2 cents on testing from a CD install of Puppy 2.

Dewdrop

System Info:
17” AOC Monitor
Processor: Intel Pentium 4
CPU 3.4Ghz
Total RAM: 904920kb
HDD: IDE
Sound: HDA-Intel - never recognized by Pup2, but works fine in 1.0.7
(mis-identified by Pup2 as azx 00:1b.0, Class 0403 8086:2668 (Rev 3)

For Puppy 1.0.7

usr_devx.sfs and usr_more.sfs are mounted on same location as pup001.

For Puppy 2

I tested the following – used the xorg wizard to test each one:

1. Screen Resolutions
640x480x16;
800x600x16;
1024x768x16 (selected by default mechanism at start of test);
1280x1024x16;
1600x1200x16

a.I did not test the x24 versions, although they were noted as being supported.

b.I would have had to manually adjust the monitor's screen size to center the desktops but it would not have needed much adjusting.

c.The default pawdesk image was present at each resolution.

2.Screen Icons
The only icon that did not respond was the Contacts icon.
Each of the others responded in a normal fashion.

3.Sound
Although there was no sound with Gxine, it did play the video of an Eagle's Concert in Australia called “Eagle's Farewell 1 Tour – Live from Melbourne." No date found on any of the DVD info..so don't know how old the DVD is.

4. Internet
DCHP and Firewall installed with no problems.
Seamonkey goes to the Puppy Home Page, and Blinky works as expected.

5. I did not attempt to change the default window manager.

[Guest]

What about booting from live-CD on your PC, is that working?

no, it's exactly the same as a poor man's install

df /initrd/pup_rw:

Filesystem 1k-blocks Used Available Use% Mounted on
tmpfs 0 0 0 0% /initrd/pup_rw

[GuestToo]

forgot to sign in

[J_Rey]

1: OK the problem with JWM 1.4 being much slower than 0.24 under QEMU has been reported to the developer of JWM (via chatroom). He didn't immediately figure out the reason for the slowness, but was working on figuring it out.

2: My BIOS is from the year 2000 and I flashed it to the latest with ACPI support. When I don't provide a acpi= boot parameter then I get this:

Code: Select all

ACPI is disabled because your bios is from 2000 and too old
You can enable it with acpi=force

So not providing an ACPI parameter auto detects (kinda) for ACPI support. I can use Puppy successfully with all three methods.

3: I always boot Puppy Linux from a DOS prompt with either Gujin (tiny.exe) or linld.com from my hard drive. So, I found the same error with the PMEDIA boot parameter mentioned earlier. But while it was stopped I decided to scroll up (Shift-Page Up) to see if there were any other boot errors/messages and I found some interesting things (non-interesting edited out):

Code: Select all

loop: loaded (max 8 devices)
RAMDISK driver initialized: 16 RAM disks of 12288K size 1024 blocksize
kmod: failed to exec /sbin/modprobe - s -k scsi_hostadapter, errno=2
/sbin/init: 402: probedisk: not found
TEST PUPMODE=1 PUPSAVE=
ERROR, cannot find Puppy on idehd boot media.
PUPMODE=1  PDEV1=
Kernel panic: Attempted to kill init!

So, with no PMEDIA parameter and PMEDIA= I got the above error. PMEDIA=idehd worked as expected, but I expected Puppy 2 to autodetect where it was located! Also, the directory /mnt/home was there but blank and confused me for a minute.

4: The ALSA config at startup worked in X but when I selected Yes to play the sample, it never played it just skipped it and went to X. (?!?!?)

5: Great news! XOrgWizard in Puppy 2a and 1.0.8r1 both work as they should for my video chipset, Intel 82810-DC100 (i810)!

6: I much prefer SeaMonkey over the other browsers/suites. But I noticed that you didn't compile with Roaming Profile support! That would be a great thing to add, so I could share my profile from my Windoze install. Also, for the final release, all the references to Mozilla should be replaced with SeaMonkey (e.g. in the menu and no symlinks, etc.).

7: What will the extra addon files be expected to be named? E.g. instead of only usr_more.sfs, would it be xtra_123.sfs or ooo_202.sfs?

8: Of course the Help files will need revamped, but also I'd like to suggest mirroring them online.

[Lobster]

6: I much prefer SeaMonkey over the other browsers/suites. But I noticed that you didn't compile with Roaming Profile support! That would be a great thing to add,

I always have to change preferences / appearance / fonts
from serif to san serif (16) if this can be compiled in it makes all browsing more ledgible . . .

[BarryK]

3.
I think that that latest puppy2 has fixed those errors. So, not having any
PMEDIA, puppy should now be able to locate the files.

4.
I think that is fixed now.

7.
The .sfs files are now versioned, that is, for puppy version 2.0.0, all the
.sfs files must have "_200.sfs". This can be prefixed with anything you
like.
pup_200.sfs is reserved as this is puppy himself.
devx_200.sfs I'm using as the development/compiler environment.
So, apart from those two, you can have anything_200.sfs.

They are expected to be located at the same place as the pup_save.3fs file.

Here is the code that loads them:

Code: Select all

 CNTLOOP=3
  for ONESFS in `ls -1 *_${PUPPYVERSION}.sfs | grep -v "^pup_" | tr "\n" " "`
  do
   losetup-FULL /dev/loop${CNTLOOP} $EXTRASFS/$ONESFS
   mount -r -t squashfs -o noatime /dev/loop${CNTLOOP} /pup_ro${CNTLOOP}
   [ $? -eq 0 ] && UMNTRO="${UMNTRO}:/pup_ro${CNTLOOP}=ro"
   CNTLOOP=`expr $CNTLOOP + 1`
  done

and a bit further down:

Code: Select all

 echo -n "Creating unionfs..."
 mount -t unionfs -o dirs=${UMNT1}${UMNTRO} none /pup_new;check_status $?

EXTRASFS is the mount point of the partition that has pup_save.3fs.
The only thing loaded prior to the above code is pup_200.sfs, so the rest,
including devx_200.sfs, will be loaded in alphabetical order.

[J_Rey]

OK thanks. I'm now wondering how you plan to ensure that say an OpenOffice.org 2.0.2 SquashFS file has future and backward compatibility? Renaming the file seems ackward, but changes between releases might cause errors also. Hmm....

Also, according to the boot message I quoted above there is a maximum of 8 loop devices. So, to trap potential unrecognized errors, with ^Cricket^/Anacari's scripting help, I recommend something like:

Code: Select all

   if CNTLOOP=8 then `echo ERROR: maximum amount of loop devices reached` fi

before

Code: Select all

   CNTLOOP=`expr $CNTLOOP + 1`

[J_Rey]

I just remembered! Is there already or could we have the ability to add the contents of *.sfs files when remastering a new CD image? That would be great if implemented!

[BarryK]

I just remembered! Is there already or could we have the ability to add the contents of *.sfs files when remastering a new CD image? That would be great if implemented!

Interesting question... in the case of the multisession CD/DVD, that would
already work, as the CD/DVD is where the save folders are. So, the above
code would look on the CD/DVD -- but then, it would have to stay mounted.

Yes, in general, I think it is feasible. If *.sfs files are found on the boot media,
they can be copied to the partition that has the pup_save.3fs file ...hmmm, I
may have already implemented this, or was that back on puppy1?
Anyway, will check it out.

[Marv]

3 or so weeks on puppy2a. Very smooth for a pre-beta. This is a syslinux install to a cf card on an ide-cf adapter. Intel 815 mobo, PIII 667, 512 Mb RAM, ATI rage 128 video.

A few comments and a question:

Xorg works well.

I set height=20 in the jwm configuration file. Better looking menu.

I had some problems with the machine being locked up or veeery slow after waking from sleep. Disabling sleep and suspend in gxset seemed to fix this. Probably an ACPI issue. The same machine ran fine with the default sleep and suspend under puppy-1.0.7.

Installing alien apps is much smoother than under 1.0.x. Opera 9 installed perfectly using the standard opera install.sh.

The question:

On my system, X memory usage rises as the system is used longer and eventually reaches 80 or 90% of available and the system slows to a crawl. This seems to be true for Seamonkey use as well as Opera and under puppy2a does not return to a few percent when apps are shut down as it does under 1.0.7 on the identical machine.

Could someone try the following in puppy2a?
Fresh boot, look at top processes. (my X mem use 1.7%)
Browse, use machine and recheck X memory use as above over time when apps are open and after shutting them down.

Thanks

[BarryK]

On my system, X memory usage rises as the system is used longer and eventually reaches 80 or 90% of available and the system slows to a crawl. This seems to be true for Seamonkey use as well as Opera and under puppy2a does not return to a few percent when apps are shut down as it does under 1.0.7 on the identical machine.

Could someone try the following in puppy2a?
Fresh boot, look at top processes. (my X mem use 1.7%)
Browse, use machine and recheck X memory use as above over time when apps are open and after shutting them down.

Thanks

What time period are you looking at here?
I ran puppy2 for two hours, surfing the Internet, and it did seem that when shut
everything down, X did not drop as low as before. Maybe, hard to say, %
mem usage is all over the place. But after 2 hours and after quiting Seamonkey etc., it was about 3.5%. Whereas at the start it was 2.2%.

I don't recall, maybe puppy2alpha has the faulty version of Blinky, but then,
if you run top, you should see blinky climbing up quite quickly.
Easy solution, kill blinky.

[Marv]

What time period are you looking at here?
I ran puppy2 for two hours, surfing the Internet, and it did seem that when shut
everything down, X did not drop as low as before. Maybe, hard to say, %
mem usage is all over the place. But after 2 hours and after quiting Seamonkey etc., it was about 3.5%. Whereas at the start it was 2.2%.

I don't recall, maybe puppy2alpha has the faulty version of Blinky, but then,
if you run top, you should see blinky climbing up quite quickly.
Easy solution, kill blinky

The rise was over a couple of hours of Seamonkey use (to 77 or 80%). Things slowed way down at that point. I am pretty sure I had blinky killed but will have to switch back to puppy2a to check. I did reboot and replicate several times so Blinky could have been on again. The high memory % seen in Top was definitely reported as X, not Blinky. The kind of slight rise you are seeing in 2a is just about what I see in 1.0.7 and seems pretty much down in the noise.