Puppy 4 Beta Bugs

[nic2109]

I think this has been noted before (in early alpha testing), but in Dingo PUI still appears to only have eyes for Puppy 398 and won't install any other release.

When it cannot find a "398" .sfs file it bales out with the sanity check failure message.

It's true of both CD and Directory sources.

[bde]

@bde: I'm using WPA encrytion and it's a built-in Intel PRO/Wireless 3945A "card" on the motherboard. The Module it uses is ipw3945.

Manual setup works every time (but including the oddities already described) - it just doesn't get "remembered" for the next boot so must be repeated. I just select the profile (which it has remembered), load it and click 'Use this profile'.

Blinky doesn't really show much activity until the wizard has completed; but this is what you would anticipate.

One difference I note is that for you it's interface 'wlan0' whereas for me it's 'eth1' with wired being 'eth0'. I have no idea whether that would make a difference to your solution.

I did get 'wlan0' when the early dingo alphas were on the latest kernel and using the module ipl3945, but as that only catered for WEP and not WPA (among other reasons) Barry reverted to the previous kernel for the later alphas and now beta 1.

Nick: Do you have eth1wireless in /etc directory? What does it have? I think it should have

/usr/sbin/wpa_connect.sh eth1 wext

Also, please check if /etc/wpa_supplicant.conf has the right settings (if you quote the content here please don't forget to replace with dummy key). I can not think of anything else now.

[bde]

Has anyone else been able to install .pup files? Because I have the same problem as the other users who posted in this thread before me (after installing the packages that puppy says are necessary, clicking on a .pup file just makes a box pop up with the name of the file, nothing else).

I was able to install .pup files. The required packages are dotpuphandler and gettext. Please install these packages, retry installing .pup file. Please report back if it worked or not.

[nic2109]

Nick: Do you have eth1wireless in /etc directory? What does it have? I think it should have

/usr/sbin/wpa_connect.sh eth1 wext

Also, please check if /etc/wpa_supplicant.conf has the right settings (if you quote the content here please don't forget to replace with dummy key). I can not think of anything else now.

Thanks for the response. I cannot check right now as I'm at work using a different PC. Shockingly it's a Windoze box - - and I've not yet convinced them to roll out Puppy 4 beta. Can't think why .

I'll check tonight and report back.

As you and I aren't alone in having this problem I doubt that it's hardware related. One of the gurus - I haven't found the post so cannot give you the name (or the link) - did mention that the issue isn't what is saved but it is the sequence of activating stuff when booting up.

Later that same day ....

1. /etc/eth1wireless does not exist

2. /etc/wpa_supplicant.conf looks like this :-

Code: Select all

ctrl_interface=/var/run/wpa_supplicant
ap_scan=2
update_config=1

network={
	ssid="*********"
	psk=********************a19c77**********6457fcde********************
	proto=WPA
	key_mgmt=WPA-PSK
	pairwise=TKIP
	group=TKIP
}

I've replaced sensitive data with **** etc. The SSID is correct, but i have no way of knowing whether the psk value is correct as it has been encrypted! I have left some of it un-asterisked to give the sense of the content. The actual value is 9 characters and not an enormous string.

I hope this is useful and helps crack this problem.

[oli]

The following isn't really a bug, but perhaps you can improve Puppy's hardware detection:

xorgwizard doesn't recognize my hardware (onboard graphic and Eizo monitor, connected via DVI) properly. I tested XFdrake (from PCLinuxOS) and my hardware is recognized very well. Perhaps you can have a look at XFdrake and find out why this tool has a better hardware detection than Puppy's xorgwizard.

[eprv]

Hi, I am 3.01 user, very satisfied, would like to test 4 beta but the download for me (ADSL) is about 2kb/sec about 10 hours to complete and with this rate it is always corrupted or stopped during download. Any idea?

you can download Dingo beta from me:
http://puppylover.netsons.org/dokupuppy ... a_releases

Dingo, thank you for help but this didn't work either, after the number verification on your site nothing happened, so I will probably wait to the final 4.00

[nic2109]

The following isn't really a bug, but perhaps you can improve Puppy's hardware detection:

xorgwizard doesn't recognize my hardware (onboard graphic and Eizo monitor, connected via DVI) properly. I tested XFdrake (from PCLinuxOS) and my hardware is recognized very well. Perhaps you can have a look at XFdrake and find out why this tool has a better hardware detection than Puppy's xorgwizard.

Any hardware guru reading this (sorry, but it's not me) will want to know some specific details about the graphics "card". Like whether it's Intel, or nVidia, or ATi, what the model and chipset is and so on.

You could also boot up XFDrake/PCLinuxOS and (in a terminal) type

Code: Select all

modprobe -l > filename_of_your_choosing

and compare the result with the same command in Puppy using a diff utility or your very own eyeballs. My guess is that Puppy's will be a shorter list which is why you will need to be specific about your hardware so that the correct module(s) can be identified and added - if suitable.

I hope this helps, and good luck.

[Wolf Pup]

/usr/lib/mozilla should be a symbolic link to /usr/lib/seamonkey and the plugins in the /usr/lib/mozilla/plugins moved to /usr/lib/seamonkey/plugins. If not seamonkey will be looking in three places for plugins instead of the usual two.

[bde]

Later that same day ....

1. /etc/eth1wireless does not exist

2. /etc/wpa_supplicant.conf looks like this :-

Code: Select all

ctrl_interface=/var/run/wpa_supplicant
ap_scan=2
update_config=1

network={
	ssid="*********"
	psk=********************a19c77**********6457fcde********************
	proto=WPA
	key_mgmt=WPA-PSK
	pairwise=TKIP
	group=TKIP
}

I've replaced sensitive data with **** etc. The SSID is correct, but i have no way of knowing whether the psk value is correct as it has been encrypted! I have left some of it un-asterisked to give the sense of the content. The actual value is 9 characters and not an enormous string.

I hope this is useful and helps crack this problem.

Nick: Can you check the following? From the console, type the command below, hit the enter key, and see if you get wireless connection. Assuming that /etc/wpa_supplicant.conf has the correct entry, and the wireless interface is eth1.

Code: Select all

/usr/sbin/wpa_connect.sh eth1 wext

After the connection is made, run from console:

Code: Select all

dhcpcd -t 30 -d eth1

This allocates the IP number to your comuter from the router, and you should be ready to browse. If you see error message that dhcpcd is already running, you have to kill this process, and run the above command again. To kill it, have to run

Code: Select all

ps -A | grep dhcpcd

note the PID (process ID number) on the left hand side of output, and have to run

Code: Select all

kill -9 PID

substituting the PID with the number you found above. Hope this works.
Thanks.

[rerwin]

A CD remastered from a puppy with a modem does not detect that modem when booting from that CD.

The /lib/modules/firmware.dep.(kernel) file contains commented lines for each module already set up from zdrv, which can include the copying of a post-install script to /init.d, within /etc. As a result with the remastered CD, the driver modules get loaded but the script is not copied from zdrv, so the modem is "not detected".

To fix that, I added copying of the contents of /etc/init.d to the /etc/init.d directiory in the remaster, after line 348.

Code: Select all

#v3.99 post-install scripts, since /lib/modules/firmware.dep.(kernel) shows them as copied into /etc/init.d.
cp -af /etc/init.d/* /tmp/etc/init.d/   #3.99

The updated script is attached. It should be copied to /usr/sbin.
Richard

[capoverde]

Update to my previous post about having to restart X for a correct (or almost) initialization: just tried to see what happens using Xorg.

Well, nothing happens -- after a seemingly normal and successful test (at 1280x1024x32) and getting the go, the screen stays black and no key combination takes back to text mode or to reboot: total hang! The MD5 sum says the ISO file is shipshape.

Using the present LCD monitor, former (stable) releases had both Xvesa and Xorg working OK, with apparently identical behavior.

[HairyWill]

Barry
You asked for ideas for stopping the battery tray app starting on desktop machines. I haven't really followed the overall problem but I suspect it might only be happening on machines that use apm rather than acpi. You can query the apm status by doing cat /proc/apm. Here are a couple of example outputs from my laptop. The first is with the power plugged in and the second is with it disconnected

Code: Select all

sh-3.00# cat /proc/apm
1.16ac 1.2 0x03 0x01 0x03 0x09 99% -1 ?
sh-3.00# cat /proc/apm
1.16ac 1.2 0x03 0x00 0x00 0x01 97% 2 min
sh-3.00#

On my laptop it is possible to test apm behaviour by booting with acpi=off apm=on

[floborg]

Barry asked for more info about the battery app issue:

I posted the output of lsmod in an earlier post.
I do not have a /proc/acpi directory.
I do not have a /sys/module/battery directory.

On what may be a related note, I've never been able to get CPU temperature information using any of the methods described in these forums (with Puppy 3.01).

[jcagle]

Not sure that this is really a bug...

I was using the Flock web browser in Alpha 7, then went to Puppy 4 Beta...

Flock doesn't work.

first of all, libstdc++.so.5 wasn't there, so I created a symlink to the newer version that was there.

Now I get this

./flock
./flock-bin: /usr/lib/libstdc++.so.5: version `GLIBCPP_3.2' not found (required by ./flock-bin)
./flock-bin: /usr/lib/libstdc++.so.5: version `CXXABI_1.2' not found (required by ./flock-bin)
./flock-bin: /usr/lib/libstdc++.so.5: version `GLIBCPP_3.2' not found (required by ./libxpcom_core.so)
./flock-bin: /usr/lib/libstdc++.so.5: version `CXXABI_1.2' not found (required by ./libxpcom_core.so)
./flock-bin: /usr/lib/libstdc++.so.5: version `GLIBCPP_3.2' not found (required by ./libxpcom_compat.so)

Don't know that this is really a bug, but probably something flock users need to have fixed.

[kirk]

The Battery applet problem Barry mentioned in his blog is with computers that use apm. The docs included with asapm (the battery app) says this should not happen:

This estimate is either
provided by the APM daemon or is calculated by this tool
itself. When there is no estimate available the display is
disabled.

The rc.modules2 script only checks for acpi batterys and loads the needed modules.

Disciple reported on this thread:

http://www.murga-linux.com/puppy/viewto ... 2&start=30

that this only occurs with alpha7 (and now beta1), both have the 2.6.21.7 kernel, and that it did not occur with alpha6 which has the 2.6.24.

1) So the first question is for Barry. Do you still plan on using 2.6.25 for Dingo final? If so this may be a non-problem.

2) Floborg, could you post the output of cat /proc/apm? Then we can compare it to what HairyWill posted and make some kind of test for it.


Also,

Noticed that shift-delete doesn't seem to copy text from rxvt in beta1.

I find Gxine's GUI easy to use, but Xine-ui did seem to work quite well.

[disciple]

Beaten to it.

Barry, I updated the trash roxapp again, with a little bugfix and some of the additional functionality I was suggesting someone else could do.
http://www.murga-linux.com/puppy/viewto ... 998#188998

I'm guessing it won't change again in the near future, except Dan is talking about modifying it to use fltk via Murgalua.

BUT, I think you should change it back to use icons in /usr/local/apps/Trash, and have the icons as links to icons in /usr/local/lib/X11/pixmaps/
Because if you have it directly using icons in /usr/local/lib/X11/pixmaps/, like in the alpha's, then it isn't backwards compatible. Whenever anyone restores or deletes the last item in their trash, if this item was deleted in an earlier puppy, then it will still look in /usr/local/lib/X11/pixmaps/ for the empty icon.
I made the icon paths variables, so you just have to look in the top of AppRun and temp-AppRun.
EDIT - I guess this isn't a big issue, as it will get the correct empty icon if you click on the trash.

[floborg]

2) Floborg, could you post the output of cat /proc/apm? Then we can compare it to what HairyWill posted and make some kind of test for it.

Code: Select all

# cat /proc/apm
1.16ac 1.2 0x03 0x01 0xff 0x80 -1% -1 ?

[Sage]

What is the significance of the /bin/mnt: line91
message in the boot up sequence of a full install of P4beta1 ?
It never appeared before.

[Firefox]

I tried to install pups to Dingo - appeared to work for normal files with getext and dotpuphandler but not with Puppy software installer - the files can be downloaded but the menu does not appear and the rate of download is to fast for my eyes/systen, i never know if the files are ok or not - it seams the arrow disapears half way along giving the impression of success.
Can somebody please make a dot PET of PSI for Dingo please.

[vito]

Hello all :

I ran into this bug in puppy 3.01 also. Everything works fine until I try and load youtube.com, then the curser freezes, and I have to reboot. Once the computer rebooted itself when I was trying to youtube.com.

Could this be a flash problem? The problem for me showed up on
TMXXINE 3.01 and DINGO BETA.

Mny Thx,

Vito