On 14/10/2013 12:36, Christoph Thompson wrote: > Hi, > > I'm surprised that no one pointed it out to the obvious and correct fix for > this sort of situation instead > of taking the easy route and just adding a symlink or using a hack in the > runtime loader. In the build > system of the software (here LLVM) the shell scripts must be changed to be > ".in" files which are > processed at build time to generate (usually with 'sed') the final script > to be installed with the adequate > shebang line for the system for which LLVM is being built. If you do it > this way, I don't see any (good) > reason why upstream won't accept the patches. > What makes you think we didn't try? The problem is that sometimes either some projects are just a bunch of python/perl/whatever scripts that don't use autofools, or that the developers don't care about weird systems and just incorrectly say "fix your OS". François.