Arthur Loiret skrev:
On Tue, May 13, 2008 at 06:21:44AM +0200, Ove Kaaven wrote:Arthur Loiret skrev:That's going to break backports, so I'd appreciate not doing that. It was possible to build the wine package on etch without patching anything.The following commit has been merged in the master branch: commit 87ae0e26dcec4ac1cb230ec3e8c33c4f943d9642 Author: Arthur Loiret <arthur.loiret@u-psud.fr> Date: Tue May 13 00:50:42 2008 +0000 Build-Depends and Depends on ia32-libs (>= 2.5).OK, I'll revert. A workaround for ensure the ia32-libs version is to add --enable-png and co. to configure flags. Then if ia32-libs doesn't work, the configure will fail.
Perhaps, but it won't help much. Etch's ia32-libs is what it is (and it contains libpng12.so.0, but not the .so symlinks). While the goal may be to eliminate the amd64 hack for lenny, we're just going to have to keep it for etch backports.
And by the way, the ia32-libs changelog that fixed #479662 didn't mention a symlink for libdbus (which I had also requested in that bug)? Was it added anyway? (Wine needs it to be able to detect certain events, such as inserting and ejecting CDs.)You are right, libdbus-1.so is missing, so is libodbc.so.
Well, the odbc is not important, but I think the dbus is probably important enough that I'll want to keep the amd64 hack until it's fixed... (or wine's configure could be patched to work around this, I suppose, but that doesn't sound like a fun thing to do.)