On Tue, 2008-08-19 at 16:14 -0500, Bob Friesenhahn wrote: > On Tue, 19 Aug 2008, Ralf Wildenhues wrote: > > IMVHO any kind of repetition of knowledge encoded in rpm and apt > > (or system vendor X package) databases is not tolerable, if it needs to > > be hand-maintained in any way. It's just too much data, getting out of > > date, and too many distributions. > > I don't understand. Wikipedia only knows about 242 Linux > distributions. What's the big deal? Well, that would be 242 possible names for 'libfoo', with an exponentially higher number of unseen dependencies. For instance, distro xyz and zyx do not ship with libfoo installed. However, distro xyz _does_ ship with all of the packages that libfoo depends upon, where zyx does not. Yet, they are both derived from distro abc. So, really, one can only hope to teach configure to (accurately) complain about the top level dependency. This is contingent upon being able to figure out what name to feed the resident packager in order to get libfoo in the first place. It really is a can of worms, I've been working on a 'sane' way to do it for several months with very little progress. > It seems that the world would be a simpler place if the world could > standardize on just one distribution such as the one called "Windows > Vista". Then we would not need Autoconf. LSB seems to be aiming to do that, however it is not getting a very warm reception outside of the rpm world. I'm not particularly fond of parts of it myself, but its a work in progress. Cheers, --Tim -- Monkey + Typewriter = Echoreply ( http://echoreply.us ) _______________________________________________ Autoconf mailing list Autoconf@xxxxxxx http://lists.gnu.org/mailman/listinfo/autoconf
