On Fri, Sep 13, 2002 at 11:42:27AM -0400, Joey Hess wrote: > Anthony Towns wrote: > > On Thu, Sep 12, 2002 at 12:05:39PM -0400, Joey Hess wrote: > > > Genralize it past library packages and this isn't really much of an > > > option though. Something like debconf FE can break backwards > > > compatability and yet has no soname, unfortunatly. > > So why don't we give it a "soname" ? "Depends: debconf-1 (>= 1.0.24)", eg? > We're going to do just that once we rework debconf policy a bit. It will > start at debconf-2 though. > However, not all packages should do that right off, Why not? > and this new field > offers a good way to recover from lack of foresight. If you're willing to change your package name, lack of foresight isn't a problem anyway. Rather than going from "debconf-1 (>= 1.0.24)" to "debconf-2 (>= 2.0.1)", you go from "debconf (>= 1.0.24)" to the same thing. Apart from an instinctive "ewww" at cluttering the package name a bit, I can't see any benefits to a "ABI-Got-Broken-At:" header over just versioning the package name a la library packages. Doing things this way conceivably also lets you have both debconf-1 and debconf-2 available at the same time, making the transition a little easier, just like it is for regular libraries. Cheers, aj -- Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/> I don't speak for anyone save myself. GPG signed mail preferred. ``If you don't do it now, you'll be one year older when you do.''
Attachment:
pgp_hE84IzvRL.pgp
Description: PGP signature