.next alpha image


Ah cool, I like that. I'll let you know how I get on playing :)
 
the wifi led is set to blink on activity, i found that slightly more usefull than a static led that showed the same thing as the nm-applet.
In early versions of Zaxxon, the Bluetooth LED would remain lit when bluetooth was active then blink when there was activity.  Could the wifi and Bluetooth LEDs be set to do the same here?

- Neelix
 
Okay, some notes from my tests last night. I'm using the OpenBox GUI

  •  Not sure why "Switch WiFi Hardware" and "Switch WiFi Applet" are seperate, surely if you're enabling hardware, you'll want the applet so it'll connect
  • Pressing the Pandora key will always move your mouse and open the menu in the same place. Can you not leave the mouse in it's present position?
  • I flicked the power switch and it took a few seconds to enter low-power mode, but I am now unable to wake it up and had to soft reset
  • DeadBeef segfaults (see below)
  • Despite installing the Codec pack, smplayer2 would not play MP3 files (log below)
  • "Toggle Warlock Bar" is not perminant, not sure if it's supposed to be
  • Vice x64 is not producing sound using any of SDL, OSS, and Alsa drivers (log below)
  • Firefox worked fine from my brief test
  • Had to set up wireless connection with "sudo nm-applet"
  • When going into the ServiceManager to indicate I want Samba and Dropbear to start at bootup, Samba unticks itself after I click OK
  • Trying to connect to my work network, just instantly disconnects me (log below)
Deadbeef pnd.out

./scripts/deadbeef.sh: line 11: 1440 Segmentation fault cp $REAL_HOME/.Xauthority /mnt/utmp/deadbeaf
/usr/pandora/scripts/pnd_run.sh: line 522: 1439 Segmentation fault "./$EXENAME" $ARGUMENTS "$@"

smplayer2 log

/mnt/utmp/smplayer2/bin/mplayer: error while loading shared libraries: libfribidi.so.0: cannot open shared object file: No such file or directory

Vice log

Error - Unknown resource 'FFMPEGFormat'
Warning - /mnt/utmp/vice/.vice/sdl-vicerc: Unknown resource specification at line 1840, 1841, 1842, 1843, 1844, and 1845

nm-applet log

** (nm-applet:2688): CRITICAL **: nm_connection_need_secrets: assertion `connection != NULL' failed
** (nm-applet:2688): WARNING **: get_secrets_cb: error getting connection secrets: (32) No agents were available for this request.


Edit: Lunchtime test:

  • XTree fails with "xterm: command not found"
  • KeePassX segfaults (this is probably one for me)
  • QEmu seems to work okay
  • SCUMMVM seems to work okay
  • TreeSheets - click on File->New and the Okay and Cancel buttons are too thin to read any text (Is this just an OpenBox UI setting?), otherwise seems ok
Edit:Oh, and a big one, you can't seem to Alt+Tap to right click. Is this an OpenBox setting too?

Edit:

FBA errors

Code:
./capex.sh: line 26: [: missing ']'
fba.sh: line 73: [: missing ']'
./fba: error while loading shared libraries: libpng.so.3: cannot open shared object file: No such file or directory
 
Last edited by a moderator:
Hi, for Openbox-specific stuff I'll answer in this new thread here

Okay, some notes from my tests last night. I'm using the OpenBox GUI

  • I flicked the power switch and it took a few seconds to enter low-power mode, but I am now unable to wake it up and had to soft reset
  • When going into the ServiceManager to indicate I want Samba and Dropbear to start at bootup, Samba unticks itself after I click OK
  • DeadBeef segfaults (see below)
  • Despite installing the Codec pack, smplayer2 would not play MP3 files (log below)
  • Vice x64 is not producing sound using any of SDL, OSS, and Alsa drivers (log below)
  • When going into the ServiceManager to indicate I want Samba and Dropbear to start at bootup, Samba unticks itself after I click OK
Deadbeef pnd.out

./scripts/deadbeef.sh: line 11: 1440 Segmentation fault cp $REAL_HOME/.Xauthority /mnt/utmp/deadbeaf
/usr/pandora/scripts/pnd_run.sh: line 522: 1439 Segmentation fault "./$EXENAME" $ARGUMENTS "$@"
smplayer2 log


/mnt/utmp/smplayer2/bin/mplayer: error while loading shared libraries: libfribidi.so.0: cannot open shared object file: No such file or directory

Vice log

Error - Unknown resource 'FFMPEGFormat'
Warning - /mnt/utmp/vice/.vice/sdl-vicerc: Unknown resource specification at line 1840, 1841, 1842, 1843, 1844, and 1845


Re: suspend. This was partly true for me too - after suspending, it wouldn't wake up when pressing a key, I had to press the power switch again for it to come back.

Re: samba. It seems like the script is missing. Service Manager mostly used the 'init' style of configuring the startup, which has been semi-replaced by systemd. I don't know if the goal is to go 'fully systemd' - if so, them things like Service Manager probably need re-doing.

Re: Deadbeef. It'll probably need re-PND'ing (like lots of others, too). I think aTc mentioned that even stuff like the 'cp' command can cause segfaults because of all the new libraries in .next.

Re: smplayer2 / codec pack. The codec pack mostly provided gstreamer codecs which are 1) very outdated, and 2) not used by smplayer (it uses libav - a 'rival' backend). libfribidi isn't currently compiled for .next, so unless that changes, the PND will need re-doing to include it.

Re: vice. Seems like an upgraded library problem too ...
 
Last edited by a moderator:
I think aTc mentioned that even stuff like the 'cp' command can cause segfaults because of all the new libraries in .next.
Sure if libraries like libc are included in the PND, cp is likely to segfault (linked with a more recent version of that lib than what it is given at runtime...).
As the pnd_run.sh maintainer, I've decided to trust the PND maintainers. If there is libraries in the PND, I force the system to use them. That's particularly usefull when you want libraries in a newer or patched version (like the SDL from notaz). Including libraries from superzaxoon in a PND is one of the best way to be sure your PND will fail in .next. Adding libc, libdl, librt, libacl, libattr or any core libraries in a PND is a recipice for failure.

Luckily, anyone can repack a PND with these removed, even if the maintainer is out of the place now.
 
sebt3,

this gives me an idea of a potential way of allowing to give older PNDs some more compatibility with .next, if you make pnd_run.sh to be comfigurable with a toggle, that the user can decide, if packed libs should be used, if system libs of the same name are present, too?

Default="yes" for SuperZaxxon, but in .next, that default could be set to "no". This might improve compatiblity for some PNDs.
 
Or maybe even use the PXML "minimum version" to see if it's designed for SuperZaxxon.

One thing that is bugging me, and it might be unfounded, but all the additional libraries ptitSeb added into the Code::Blocks PND, will/can these cause conflicts? How will I know what libraries need including in the PND? or should I just opkg install all the dev tools onto my system? That way, any lib that's missing and need to build I know must be included?
 
I'm unsure, anyone want to wait for pnd_run.sh to check every single libraries in a PND to check if there is an updated version in the OS.

I already removed some nice features (advanced error detection, error line marking, and log timestamps) from that script for performance reason.

Beside, I can add a directory or remove it from a list. I've no way to tell use this lib, but not this one in the same directory.

Yet a partial implementation would have a significant impact on a PND startup (as if a few mknod, a loopback creation and 2 mounts arent already enough to slow down PND startup).

If the library come from superzaxon, why would a PND maintainer include it in the PND is way over my understanding abilities. There not a single good reason to do so and a large brunch of reason to not to. If dont want to trust the system libraries, provide static libraries. plain and simple. Autotools (configure script) have a --static flag to do so. Way easier and much more reliable. It's not as if nobody does that. Every single loki release are static binaries. in fact 90% of the commercial program for linux are static binaries, like acrobat and so many more. The other 10% is a maintenance hell. At least, Oracle allow you to relink the binaries using your system libraries...

@PMProg. Last time I checked, audacious start on .next after a few tweaks here and there, yet I include a large brunch of libraries in the PND. These libraries have been built on top of zaxon lib. So it works there and it should work on .next as long as the sytem libraries these libs use are backward compatible. As long as you dont provide any system libraries in your PND, you should be fine.
 
If the library come from superzaxon, why would a PND maintainer include it in the PND is way over my understanding abilities.
I have an awser for that. The script I provide in code::blocks PND, called "copy_libs.sh" is a dumb script to copy all lib that are used by a software and not in /lib or /usr/lib. I developped this script because I have been hit a few times by the "the PND doesn't start" a few times.

The problem is, in the code::blocks PND, I have a few symlink of some librairies. The symlink links to the superzaxxon lib, and I needed that to have an easy and transparent flags for the linker.

The problem is "copy_libs" is a so dumb script that it copies the symlinks libs, thinking there are not in superzaxxon firmware. Thus, when you "copy_libs", your PND is running well under superzaxxon, but you carry useless libs (and now bad libs), like libdl, libts or libEGL to name a few.

I should update copy_libs to not copy symlinked libs, or something like that.

here is the sources of copy_libs:

Code:
#!/bin/bash

if [ -z "$1" ]
then
     echo "Usage $0 exefile

Copy all so dynmic libraries files not in /lib or /usr/lib in curent folder
And automatically call strip for program and copied .so
"
else
        pFILE="$1"        # copy bin file libs
        files=""
    ## use ldd to get shared libs list ###
        files="$(ldd $pFILE |  awk '{ print $3 }' | sed  '/^$/d')"
        #files="$(ldd $pFILE) |  awk '{ print $3 }' )"
 
        for i in $files
        do
          dcc="${i%/*}" # get dirname only
      #dcc="$(dirname '${i}')"
          # [ ! -d ${d}${dcc} ] && mkdir -p ${d}${dcc}
#      if [ "$dcc" != "/lib" ] && [ "$dcc" != "/usr/lib" ] && [ "$dcc" != "not" ]
      if [[ "$dcc" == /mnt/utmp/codeblocks/* ]]
      then
        echo "copy ${i}"
            cp -n ${i} .
#        strip $(filename "${i}")
      fi
        done
fi
strip $1
strip *.so.*
 
Following the devtools link

http://pandorawiki.org/Compile_directly_on_the_Pandora

It tells me that gcc etc. are already installed, which was nice, it did tell me libgcc-dev could not be installed though, not sure if it's important

Edit: I've just tried to shutdown the Pandora after installing a few packages, and it's now sat with the following on the screen

Code:
{ OK } Stopped Remount Root and Kernal File Systems.
       Starting Update UTMP about System Shutdown...
{ OK } Reached target Shutdown.
Even if I flick or hold the power switch it, it won't turn off.
 
Last edited by a moderator:
sebt3 seems to have some problems with the gles stuff on his rebirth pandora.

Can anyone confirm if gles works or not on pandoras with an /etc/powervr-esrev other than 2 ?

(which should be the rebirth and 1ghz versions)

Easiest way to test is probably http://repo.openpandora.org/?page=detail&app=a7xpg_m-ht

Which runs without any problem on my pandora, with a clean install of the latest image.
 
sebt3 seems to have some problems with the gles stuff on his rebirth pandora.

Can anyone confirm if gles works or not on pandoras with an /etc/powervr-esrev other than 2 ?

(which should be the rebirth and 1ghz versions)

Easiest way to test is probably http://repo.openpandora.org/?page=detail&app=a7xpg_m-ht

Which runs without any problem on my pandora, with a clean install of the latest image.
I can't reproduce this with my rebirth pandora, as in "it runs fine".

It seems like I installed the latest image (my downloaded tar.xz has the following creation time: 2013-08-18 18:39).
 
aTc, can you add a bit more details to your first post plz, e.g. name of the archive.
 
There's only one file in there that could really be the image, but i've changed the link in the first post to be the image itself instead of the dir that contains it.

Other than that there aren't really much more details needed other than extracting it to sd, and booting it in the pandora.

It's not really meant to be a totally noob friendly experience at this stage anyway :)
 
I have a Rebirth Pandora

On my image (downloaded 21st Aug), A7XPG doesn't work. pnd out says:

Code:
object.Exception@Logger.d(25): Error: abagames.util.sdl.SDLInitFailedException.SDLInitFailedException@SDLInitFailedException.d(13): Unable to open EGL context
cat /etc/powervr-esrev returns me "3"
 
I seem to miss something as I can't get setup the latest image. :(

I've extracted it to a freshly formatted ext2 SD-Card.

It boots up and the first run wizard runs until the dialog for setting up the login name is displayed.

I can enter whatever I want but it repeats displaying this dialog forever, even if i hit cancel.

Is there any log where I can look what went wrong?

I've had successfully installed the image before this latest one...

Thanks in advance
 
I think it has to do with the touchscreen calibration, if you skip that, it should be fine.
 
I've extracted the lastest tar.gz image to a freshly formatted ext3 SD-Card.

It boots up but I can't get wizard (Booting stop on "Btrfs loaded", got 1Ghz edition).

I think I miss something.

Thanks.
 
Back
Top