Re: gcc 3.2 transition in unstable
On Tue, Jan 07, 2003 at 08:57:19AM +0100, Adrian 'Dagurashibanipal' von Bidder wrote:
> On Tue, 2003-01-07 at 07:43, Chris Cheney wrote:
> > > In the gcc-3.2 transition the main problem to not change sonames is to
> > > keep compatibility with other distributions that are not changing
> > > sonames in their gcc-3.2 transition.
> > imho the concept of sonames is broken... Any number of things can cause
> > you to need to bump the soname technically, compile with a different
> > version of a library, a different version of compiler, etc.
> The soname stuff is not broken. People just want it do things it was not
> intended for. soname purely reflects the API, there should be other
> means of distinguishing the ABI (If I understand the gcc development
> correctly, that's tha main problem they're trying to figure out in
> current development. Versioned symbols are not the thing, though, afaics
> they, too, are only related to the ABI).
Yep, people are trying to fit a square peg in a round hole (or something
like that) ;) Even libtool reinforces this idea since its method of
bumping sonames is rather nice. Too bad there isn't a standardized way
of distinguishing ABI compatibility.