[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [RFC] Breaking britney2's backward compatibility



On Fri, 2011-08-12 at 10:42 +0200, Julien Cristau wrote:
> On Thu, Aug 11, 2011 at 21:57:22 +0100, Adam D. Barratt wrote:
> > Obsolete binary packages in a configurable list of archive sections
[...]
> are not automatically removed when the source package is
> > updated; instead they are kept in testing temporarily to ease
> > transitions.
[...
> So this part worries me slightly, as it means there's a greater chance
> of having programs in testing loading two incompatible versions of a
> library.  If it could be opt-in (with a 'keep xulrunner1.9.1' sort of
> hint) I'd be a bit happier.
> Then again maybe I'm overestimating the risk of this causing breakage, I
> don't know...

Phil pointed out on IRC that we should end up in this situation in
unstable reasonably often during transitions, where people partially
upgrade before the old library packages have been decrufted, and we
don't appear to be drowning in bug reports.

otoh, maybe unstable users just assume that such issues are transition
related and don't report them.

The key question is whether we think it's enough of an issue that we
want to stay in compatible mode until we could implement something like
the opt-in model.

Regards,

Adam


Reply to: