Bugzilla Bug 121937

pouetChess is a 3D and open source chess game
(http://pouetchess.sourceforge.net). Here's an ebuild for it

Created an attachment (id=79090) [details]
ebuild for version 0.11

I prepared another ebuild for this independantly:

http://genstef.homelinux.org/local/games-board/pouetchess/pouetchess-0.11.ebuild

Neither ebuild is of portage-quality as is.

Created an attachment (id=79159) [details]
pouetchess-0.11.ebuild

Not sure about the opengl/glu stuff, this always confuses the hell out of me.
Might also want to patch for homedir-configfile-fun. Your turn :)
(untested btw due to lack of a proper gfx card)

pouetchess-0.2 is out and includes:
- a scons-based build system
- config files in home directory
- an embedded AI!

I'll attach my current ebuild for it, issues remaining:
- CFLAGS and friends are not respected...
- I get 2 sandbox violations (a scons trademark), in spite of adding a prefix
parameter...

open_wr:   /usr/include/SDL/.scons21369
open_wr:   /usr/include/SDL/.sconsign

- scons install does not respect DESTDIR, so I install files by hand

Created an attachment (id=87992) [details]
0.2.0 ebuild

Created an attachment (id=88066) [details]
0.2.0 ebuild

The correct version this time (the first one did not install the data in the
right place), sorry

Created an attachment (id=88111) [details]
new iteration of the pouetchess ebuild

new iteration of the pouetchess ebuild

additions:

- keyword ~x86 (works on my installation)
- keyword ~ppc (not testet)
- patch for sandbox violations due to scons
- desktop entry

Created an attachment (id=88112) [details]
patch for the SConstruct file

this adds two lines to the SConstruct file, as seen on

http://permalink.gmane.org/gmane.comp.audio.csound.devel/6646

some comments:

make the variables at the top in the same format as the other games ebuilds:
blank line after SRC_URI, MY_PN at the top, sort KEYWORDS, RDEPEND first and
mentioned in DEPEND

patch in src_unpack, not src_compile

don't install COPYING

Created an attachment (id=88468) [details]
pouetchess-0.2.0.ebuild

cleaned up as mr.bones advised

# emerge -v pouetchess
Calculating dependencies... done!
>>> Emerging (1 of 1) games-board/pouetchess-0.2.0 to /
>>> Creating Manifest for /usr/local/portage/games-board/pouetchess
>>> Unpacking source...
>>> Unpacking pouetchess_src_0.2.0.tar.gz to /var/tmp/portage/pouetchess-0.2.0/work
 * Applying sconstruct-sandbox.patch ...                                  [ ok
]>>> Source unpacked.
>>> Compiling source in /var/tmp/portage/pouetchess-0.2.0/work/pouetchess_src_0.2.0 ...
scons: Reading SConscript files ...

Now configuring.  If something fails, consult `config.log' for details.

Detected platform : linux
Checking gcc version... 4.1.1
profiling NOT enabled, debugging NOT enabled, level 1 optimizing enabled
Detecting processor...
Couldn't detect your CPU, guessing i686 compatible..
Checking for SDL...
  Checking for sdl-config... /usr/bin/sdl-config
  Checking for LibSDL >= 1.2.8... 1.2.10
You need LibSDL version 1.2.8 or greater for this program

!!! ERROR: games-board/pouetchess-0.2.0 failed.
Call stack:
  ebuild.sh, line 1531:   Called dyn_compile
  ebuild.sh, line 931:   Called src_compile
  pouetchess-0.2.0.ebuild, line 46:   Called die

!!! scons configure failed
!!! If you need support, post the topmost build error, and the call stack if
relevant.




It looks like it could use some work.  Feel free to REOPEN this if there's a
newer upstream release or a patch for this.

I had a fix for this one in the first 0.2.0 ebuild. I'll attach a merged one

Also, does somebody compile with nvidia opengl? I encountered problems until i
switched to xorg-x11 (like suggested for celestia), so I put a pkg_setup
function with some einfos on it (same as celestia). If you do not have the
problem, please disregard it from the ebuild

EDIT: hum, cannot reopen, I'll just attach the ebuild then ;)

Created an attachment (id=88554) [details]
Ebuild with libsdl detection fix

Cool.

I'll look into it.  Thanks for the update.  =]

OK.  It compiles on my box fine, but only if I do the eselect junk.  We really
should be patching the problem rather than masking it.  Asking users to change
their opengl provider to compile something isn't acceptable for us anymore.

(In reply to comment #16)
After digging around a little bit, other projects got this problem with nvidia
OpenGL headers ( Wine for example at
http://www.winehq.org/pipermail/wine-patches/2005-September/021035.html ). 
The often used fix is to manually #define GL_GLEXT_PROTOTYPES *before*
including gl.h (or SDL_opengl.h in this case).

The attached patch fixes the problem for pouetchess

Created an attachment (id=88989) [details]
Fix for nvidia opengl headers

Added to CVS... thanks to all of you for the work on this.