On Wed, Jan 16, 2013 at 06:47:01PM +0000, Neil Williams wrote: > A different - similarly strict - criterion would catch out glib2, gtk > and others- introspection / marshalling code. This can cause a build > failure but can also cause more difficult bugs which only show at > runtime. Skipping the execution causes missing functionality, copying > data from the build architecture is likely to cause other kinds of > misbehaviour or crashes.
Yeah, that and gtk-doc. I do have some ideas on how to deal with gobject-introspection; but I think it's going to require a gobject-introspection-TRIPLET for each host architecture, which is one reason I want to have a reasonably generalisable solution to the build-dep substitution problem. There's probably no general magic bullet to the problem where cross-built packages build successfully but turn out to be inadequate in some way, although I guess that file list comparison against natively-built packages would catch a majority of the problems. -- Colin Watson [cjwat...@debian.org] -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20130116215218.ga24...@riva.dynamic.greenend.org.uk
