On Thu, Oct 06, 2005 at 02:23:53PM -0700, Philip Lamb wrote: > Wouldn't a better solution just be to make sure that cctools is > listed as a BuildDepends for every package that needs the compiler, > etc. to build (not sure whether there would be many exceptions to > this!),
That is part of what we just implemented:) Well, it's an automatic (implicit) thing, so fink itself handles it instead of actually patching all those .info files, but anyway. > and as a Depends: for the few packages that need things like > make in their postinstall phase (scilab has already been mentioned.) Ooh, forgot about that. Maintainers would have to make that explicit. OTOH, I wonder if some of those cases could be solved by doing [whatever] during the package building phases (may take some hacking). > The other problem.. fink not being able to do a selfupdate would also > be solved this way.. make fink BuildDepends: cctools. That leads to the same problem we originally had: selfupdate (via rsync or cvs) detects there is a new fink present so tries to compile it. In the original case, the build crashes if there's no compiler. In your example, fink would crash because it couldn't satisfy the BuildDepends. In both cases, fink is kinda wedged for all future selfupdate attempts. I don't see anything gained here, except perhaps a nicer error message each time the wedge hits. dan -- Daniel Macks [EMAIL PROTECTED] http://www.netspace.org/~dmacks ------------------------------------------------------- This SF.Net email is sponsored by: Power Architecture Resource Center: Free content, downloads, discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl _______________________________________________ Fink-devel mailing list Fink-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fink-devel
