2008/3/30, Duncan <address@hidden>:
Beso <address@hidden> posted
address@hidden, excerpted
below, on Sun, 30 Mar 2008 11:04:23 +0000:
> well this wouldn't be difficult to fix in gentoo since the ebuild could
> see what version is installed and apply a patch for fixing the includes
> based on the installed version of glib. the other distros might need
> another way to fix this up, though.
Imagine meeting you here! =8^)
Yes, Gentoo can quite easily (almost trivially, now that the parameters
are known) fix it with ebuild wizardry, no problem there. Similarly the
various binary distributions, most of which after all only support a
relatively narrow version range on actively supported releases, building
everything to what they ship with and perhaps supporting the next stable
during their support life.
The problem is that ideally, distributions support upstream, and upstream
ideally wants to support the widest version range possible. Creating a
patch not just usable for a particular distribution, or a particular
distribution's version range, but for the widest range possible, now
/that/ is a rather less trivial task, but it's the one we've set out to
accomplish, therefore best helping not just one distribution, but
upstream, and with it, as wide a range of existing and upgrade versions
as reasonably possible.
other distros have usually precompiled binaries, and thus with known glib version, so once the problem is identified and known i don't think that should be difficult for them to prebuild the right stuff. also, their installers usually just installs the stuff and don't need to compile it. so this issue doesn't have a real impact on the normal users of the other distros, while it does on gentoo ones.
maybe the simplest thing is to search for the includes and if not found use the pan defaults. this shouldn't be very difficult and would release the problem with the use of some #ifdefs or #ifndefs.
but i agree with you that having something to work on the major number of distros is better than having it work just on a few.
--