[SOLVED] KS fails on splash screen -- Arch Linux 64

[dragonfi]

I know how off-putting hard-to-reproduce bugs can be, so I would like to note first that the game is really marvelous, and ran without errors for me in the past.

When I start the game, it flashes the splash screen, then immediately dies.
(In the past, the game worked just fine, so I suspect some new updates may have caused it.)

Versions:
SDL 1.2.14
pygame 1.9.1

Tested with Act1 v5 and v4.
It is only Katawa Shoujo, I can play other ren'py-based visual novels just fine.
The demos from ren'py 6.10.2 runs fine. (If I'm right that is the version KS modified.)

Terminal output:

Code: Select all

$ ./Katawa\ Shoujo\ Act\ 1.sh 
Traceback (most recent call last):
  File "./Katawa Shoujo Act 1.py", line 132, in <module>
    renpy.bootstrap.bootstrap(renpy_base)
  File "/home/dragonfi/opt/Katawa Shoujo Act 1 v5-linux-x86/renpy/bootstrap.py", line 242, in bootstrap
    renpy.import_all()
  File "/home/dragonfi/opt/Katawa Shoujo Act 1 v5-linux-x86/renpy/__init__.py", line 61, in import_all
    import renpy.display.scale # Must be before module.
  File "/home/dragonfi/opt/Katawa Shoujo Act 1 v5-linux-x86/renpy/display/scale.py", line 31, in <module>
    import _renpy_font
ImportError: /home/dragonfi/opt/Katawa Shoujo Act 1 v5-linux-x86/lib/linux-x86/lib/python2.5/_renpy_font.so: undefined symbol: SDL_RWFromFile
Exception exceptions.AttributeError: "'NoneType' object has no attribute 'error'" in <bound method Popen.popen_del of <subprocess.Popen object at 0xf701efcc>> ignored

$ Traceback (most recent call last):
  File "./Katawa Shoujo Act 1.py", line 132, in <module>
    renpy.bootstrap.bootstrap(renpy_base)
  File "/home/dragonfi/opt/Katawa Shoujo Act 1 v5-linux-x86/renpy/bootstrap.py", line 170, in bootstrap
    renpy.display.presplash.show(options.presplash)
  File "/home/dragonfi/opt/Katawa Shoujo Act 1 v5-linux-x86/renpy/display/presplash.py", line 101, in show
    sys.stdout.flush()
IOError: [Errno 32] Broken pipe

[Esa94]

Can confirm, this happens for me as well now.

[delta]

KS is not really supported on Linux since there's nobody on the dev team to test it, we just release them because they're there and it's easy. And this does look like a problem with Ren'Py, distro-specific on top of that (unless other people have more details). I can only recommend to run it in a VM with a compatible version of Linux.

[dragonfi]

I done some further testing, and the game starts with a stock 6.10.2 engine.
(I assume it would fail at a point when it needs some of the advanced functions.)

It also doesn't have problem with Ubuntu 11.04 64bit. (Will update next week, to see if it will cause trouble, probably.)
versions:
libsdl1.2 : 1.2.14

My next guess would be checking the contents and the linkage for "_renpy_font.so". Of course, since I'm not a developer, and the shared object is a binary, there is not much I can do.

Can confirm, this happens for me as well now.

Just for the record, are you an Arch user too? (Otherwise, please specify the distro, architecture and the version numbers.

Edit: actually, would KS prefer it's own shared files over system-wide ones? (Yeah.. I just noticed that it's much more self-contained than I assumed. It even has it's own interpreter, if I understand correctly.)

[dragonfi]

The last Arch Linux update solved the problem on my machine. I don't know which libraries were the culprit, but it was probably a bug in one of the packages.
(I don't really know the internals, but updating lib32-glibc and python-sip might or might not had anything to do with it.)

It probably won't be useful to anyone, but for historical purposes, here are the updates that I done between tests:

Code: Select all

[2011-10-30 21:28] Running 'pacman-color -Sy'
[2011-10-30 21:28] synchronizing package lists
[2011-10-30 22:32] Running 'pacman-color -S -u'
[2011-10-30 22:32] starting full system upgrade
[2011-10-30 22:33] upgraded dia (0.97.1-4 -> 0.97.1-5)
[2011-10-30 22:33] upgraded dvdstyler (1.8.4.3-1 -> 2.0-1)
[2011-10-30 22:33] upgraded libpurple (2.10.0-2 -> 2.10.0-3)
[2011-10-30 22:33] upgraded finch (2.10.0-2 -> 2.10.0-3)
[2011-10-30 22:33] upgraded tzdata (2011k-1 -> 2011m-2)
[2011-10-30 22:33] Generating locales...
[2011-10-30 22:33]   en_US.UTF-8... done
[2011-10-30 22:33]   en_US.ISO-8859-1... done
[2011-10-30 22:33]   hu_HU.UTF-8... done
[2011-10-30 22:33]   hu_HU.ISO-8859-2... done
[2011-10-30 22:33] Generation complete.
[2011-10-30 22:33] upgraded glibc (2.14-6 -> 2.14.1-1)
[2011-10-30 22:33] upgraded gcc-libs (4.6.1-4 -> 4.6.2-1)
[2011-10-30 22:33] upgraded gcc (4.6.1-4 -> 4.6.2-1)
[2011-10-30 22:33] upgraded gcc-fortran (4.6.1-4 -> 4.6.2-1)
[2011-10-30 22:33] upgraded gnutls (3.0.4-2 -> 3.0.5-1)
[2011-10-30 22:33] upgraded icu (4.8.1-1 -> 4.8.1.1-1)
[2011-10-30 22:33] upgraded libqzeitgeist (0.7.0-1 -> 0.8.0-1)
[2011-10-30 22:33] upgraded phonon (4.5.55-1 -> 1:4.5.1-2)
[2011-10-30 22:34] upgraded kdelibs (4.7.2-2 -> 4.7.2-4)
[2011-10-30 22:34] upgraded lib32-glibc (2.14-6 -> 2.14.1-1)
[2011-10-30 22:34] upgraded lib32-gcc-libs (4.6.1-4 -> 4.6.2-1)
[2011-10-30 22:34] upgraded libltdl (2.4.2-1 -> 2.4.2-2)
[2011-10-30 22:34] upgraded lib32-libltdl (2.4.2-1 -> 2.4.2-2)
[2011-10-30 22:34] upgraded lib32-pcre (8.12-1 -> 8.20-1)
[2011-10-30 22:34] upgraded libtool (2.4.2-1 -> 2.4.2-2)
[2011-10-30 22:34] upgraded pam (1.1.4-1 -> 1.1.5-1)
[2011-10-30 22:34] upgraded pidgin (2.10.0-2 -> 2.10.0-3)
[2011-10-30 22:34] upgraded sip (4.12.4-1 -> 4.13-1)
[2011-10-30 22:34] upgraded pyqt (4.8.5-1 -> 4.8.6-1)
[2011-10-30 22:34] upgraded python2-sip (4.12.4-1 -> 4.13-1)
[2011-10-30 22:34] upgraded python2-pyqt (4.8.5-1 -> 4.8.6-1)
[2011-10-30 22:34] upgraded sudo (1.8.3-1 -> 1.8.3.p1-1)
[2011-10-30 22:34] >>> texlive: updating the filename database...
[2011-10-30 22:34] mktexlsr: Updating /etc/texmf/ls-R... 
[2011-10-30 22:34] mktexlsr: Updating /usr/share/texmf/ls-R... 
[2011-10-30 22:34] mktexlsr: Updating /usr/share/texmf-dist/ls-R... 
[2011-10-30 22:34] mktexlsr: Updating /var/lib/texmf/ls-R... 
[2011-10-30 22:34] mktexlsr: Done.
[2011-10-30 22:35] recreating all formats...Error: `eptex -ini  -jobname=eptex -progname=eptex *eptex.ini' failed
[2011-10-30 22:35] 
[2011-10-30 22:35] ###############################################################################
[2011-10-30 22:35] fmtutil: Error! Not all formats have been built successfully.
[2011-10-30 22:35] Visit the log files in directory
[2011-10-30 22:35]   /var/lib/texmf/web2c
[2011-10-30 22:35] for details.
[2011-10-30 22:35] ###############################################################################
[2011-10-30 22:35] 
[2011-10-30 22:35] This is a summary of all `failed' messages:
[2011-10-30 22:35] `eptex -ini  -jobname=eptex -progname=eptex *eptex.ini' failed
[2011-10-30 22:35]  done.
[2011-10-30 22:35]  (logs are under /var/lib/texmf/web2c/<engine>/<formatname>.log)
[2011-10-30 22:35] upgraded texlive-bin (2011.1-2 -> 2011.1-3)
[2011-10-31 09:36] Running 'pacman-color -Sy'
[2011-10-31 09:36] synchronizing package lists
[2011-10-31 09:37] Running 'pacman-color -S -u'
[2011-10-31 09:37] starting full system upgrade
[2011-10-31 09:37] upgraded libatasmart (0.17-1 -> 0.18-1)
[2011-10-31 09:37] upgraded p7zip (9.20.1-5 -> 9.20.1-6)

[delta]

That's good to know, thanks.

[dragonfi]

Actually, Arch being a bleeding-edge distro means that it sometimes experiences bugs that, left untreated, would filter down to more mainstream distros like Ubuntu, that are usually lags behind a bit. It's good that it went away, that means it probably won't happen in Ubuntu.

I'm marking the thread as solved. (It's simply just prepending [SOLVED] to the first post, right?

[Esa94]

Just for the record, are you an Arch user too? (Otherwise, please specify the distro, architecture and the version numbers.

I know I'm late but yeah. I thought I'd posted something here about how Arch is a bleeding-edge distro but apparently not. Maybe I'm mixing things up and was instead talking about it to someone. Arch64 here as well.
Let's see if I can confirm it works... Yeah.