* Peter O'Gorman wrote on Wed, Feb 13, 2008 at 11:49:26PM CET: > Ralf Wildenhues wrote: > > Argh. AC_LANG_PROGRAM does not produce a meaningful program for Java, > > as Autoconf has no decent Java support. Gah! And we test for RUNPATH > > using $OBJDUMP in each tag, GCJ last, so there it does not get set to > > yes, and shlibpath_overrides_runpath is not a tagged variable. What a > > mess.
> The dynamic linker checks are called for every tag, but produce no > tagged variables. I was going to propose something like the attached > patch at some point, but thought it best to wait 'til after 2.x is > released in case it broke something.... maybe should not have waited. Yes, I think that has to wait. For now, I'd prefer backing out the shlibpath_overrides_runpath detection using objdump at all. But maybe I can get gcj to have a decent source file to eat; after all, ATM this is a bug only for packages using the GCJ tag. That would be much safer. We can look at your patch post 2.2. There are lots of nitty gritty ordering issues to check. :-/ Cheers, Ralf _______________________________________________ http://lists.gnu.org/mailman/listinfo/libtool
