On Wed, Nov 30, 2005 at 01:15:43AM -0500, [EMAIL PROTECTED] wrote: > >I'm afraid that applying this patch upstream *does* have a portability > >impact; the patch will do the right thing on systems that use the GNU > >linker > >or a linker with similar features, but there are other Unices that don't > >support dynamic linking or whose library format doesn't store dependency > >information, so on these systems a hardcoded -lImlib2 won't work. Also, > >not > >all GNU/Linux systems put libImlib2 in the linker search path, so > >imlib2-config --libs may output -L values needed for linking.
> I admit I have forgotten about local installation and such, which was > dumb. As for static linking, I had a good look at enlightenment CVS, and > this seems never to be supported anyway by imlib2-config, as you > suspected, so it is not a direct issue (I do not restrain nor extend > portability by ignoring the case -- it's a known limitation). Hmm, I'm rather surprised at the notion that static linking isn't supported; the build targets certainly build a static version of libImlib2 by default. > From your comments, things are now done a little differently[1]: I added > code to first try to link a demo Imlib2 program with the "-lImlib2" > switch, then I fall back reusing the output of "imlib2-config --cflags > --libs" when it doesn't work. This way, people with a ld or similar > linker and Imlib2 in their default library path will not depend directly > on outdated freetype and such as long as this is possible, while things > should continue to work without side effects for everyone else the way it > was before. Well, could still give the wrong results if the user has multiple copies of imlib installed and wants to build against a version other than the one in the default search path; but that's certainly an edge case, perhaps one not worth worrying about. My main concern is that if upstreams adopt library-handling patches that are intended to be Debian-specific and in the process break compatibility with other systems, we'll have a harder time in the long term convincing others to adopt solutions that *do* work for everyone. :) The check you describe seems to be ok in this regard. > As for the rest of your patch (AM_MAINTAINER_MODE, no X Extra, distcleans > targets, etc.), I think it is all good and should stay in too. Let me know > if you see any other problems with the updated scripts... Yours, Yeah, those bits are definitely applicable upstream. :) Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/
signature.asc
Description: Digital signature
