Comment 5 for bug 370847

Your patch uses "$SRCDIR/.." in all cases anyway. I dislike having to do the chdir() stuff, but based on the example, it looks like we have to.

As for a generic way being better, what if you wanted to use m4_esyscmd() in an argument for another m4 call that was parsed by intltool, either now or in the future. We'd have to keep adding checks like this. It would be better if all arguments to m4 calls handled by intltool, get handled in the same way in all cases. I think that is a general FIXME that can be done later though, rather than blocking this fix on it.