On Mon, 2008-10-13 at 01:26 -0400, D. Hugh Redelmeier wrote: > | From: Braden McDaniel <braden@xxxxxxxxxxxxx> > > | Yes, but the kind of patches that should be sent upstream are not the > | ones resulting in 300 busted packages from a libtool upgrade. We're > | talking about patches applied by a specfile build. > > Could you characterize the problems that cause 300 busted packages? I couldn't. I hope Karsten Hopp will (the "300" figure comes from him); he's said he's preparing a wiki page with this information. (See earlier in this thread.) > Naive questions: > > - Might a reasonable patch to libtool 2.2 fix a lot of these problems? > (And no, I don't mean "patch it back to 2.1.") I don't know. (And FYI, Fedora never shipped 2.1; I think it wasn't released. I think the 2.2 series succeeds the 1.5 one.) > - Would it make sense to allow multiple versions of libtools co-exist > on a system and allow a .spec declaration of some kind specify which > to use? That possibility has been floated; and things might even be leaning that way. See bug 459387. I don't have an objection to that approach as long as "libtoolize" (and similar) refers to the latest version of the tool--just as "autoconf" or "automake" refer to the latest versions of those tools. And as long as that's the case, specfiles for libtool2-incompatible packages that run autoreconf or libtoolize will need to make at least minor changes to request the old version. -- Braden McDaniel e-mail: <braden@xxxxxxxxxxxxx> <http://endoframe.com> Jabber: <braden@xxxxxxxxxx> -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list