Re: Minutes for release status meeting from 2010-11-15

On Wed, Nov 17, 2010 at 09:02:04AM +0100, Kay Ramme wrote:
> >Yes, but why should we do it if you could fix the library itself properly?
> >Given you so much are after compatibility a ABI breakage even without SONAMe change
> >contradicts what you yourself stand for.
> jvmaccess is _not_ part of the public / stable ABI. The breakage is
> unfortunate but not critical and does _not_ justify a delay.

Whether it's part of the public / stable URE ABI is not relevant, IMHO.
The lib has a SONAME and the ABI broke without changing it.
Reason enough to fix it properly now. Either by bumping SONAME or reverting
the change. Anything else is a bug as it violates UNIX library
principles.

(Or you made clear that people shouldn't care and make those libs not
have a SONAME and get the same style like the other internal libs)

> Go and revert the patch in your source, I ask Stephan to back it out

If it's this easy, why are you just not doing this?
I am not going to fiddle with libraries when upstream broke ABI and can fix it
himself in due course *before* the breakage gets released.
Only if it's too late (which it isn't).

(Besides that, I alreaday added the Breaks: as temporary measure in any
case).

Grüße/Regards,

René