On Sun, Jan 23, 2000 at 08:22:47AM -0500, Michael Stone wrote: > On Sat, Jan 22, 2000 at 09:23:01PM -0500, Ben Collins wrote: > > Well, I recompiled almost every package for sparc against not only > > libncurses5, but libreadline2. This took a matter of 2 days and I have yet > > to hear about any problems. Further more, because of the autobuilders, a > > lot of bug reports have already been filed concerning packages that do not > > use <curses.h> (ie. they do not check for the proper header in autoconf > > or the like). This is not as hard as you try to make it, and it is also > > already underway for some of the non-i386 ports. > > Oh, good. Someone brought up the famous proof by "it works for me." I > suppose that means that we no longer have to worry about problems with > any packages due to this change, we don't have to worry about testing > all the packages that have a major library change during freeze, and we > can generally ignore the guidelines we set up to try to maintain package > stability before we release. What a relief. > But it IS a fact that debian has had 'duplicate' libraries time and again, in at least hamm, slink, and now potato. E.g. in slink there were also two ncurses libraries (one of them, ncurses3.4, while being in oldlibs, depended on by info, psmisc, procps and ircii at least, of which the first three are quite basic unix packages I'd say), and also two sets of gtk and glib libraries, namely libgtk1/libglib1 and libgtk1.1/libglib1.1. And I wouldn't be surprised if there were other such cases as well.
Attachment:
pgpAoGIK61fQd.pgp
Description: PGP signature