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

Re: [Caml-list] Release 3.09.1



On Thu, Jan 05, 2006 at 01:04:40PM -0500, Eric Cooper wrote:
> On Thu, Jan 05, 2006 at 09:08:04AM +0100, Stefano Zacchiroli wrote:
> > On Wed, Jan 04, 2006 at 08:19:12PM +0100, Sven Luther wrote:
> > > > Unless someone objects to it, I'll package and upload OCaml 3.09.1 as
> > > > soon as 3.09.0 is in testing. Do you all agree with a full rebuild?
> > > Yes, but let's do it like this :
> > >   3) we ask Steve to requeue all those ocaml related packages for a binNMU,
> > >   and file RC bugs against all those that fail.
> > 
> > I object this way of making ocaml transitions.
> > 
> > Generally speaking a new release of ocaml may imply any kind of changes
> > including, for example, source level incompatibility. I don't want my
> > packages to be binNMU-ed for making a transition: I want to test them
> > and perform a regular sourceful upload.
> 
> I agree with the spirit of this, too, but I noticed a big difficulty
> in practice when we went through the 3.09.0 transition.  If you depend
> on lots of other OCaml components, and each one is going through this
> same process of manual inspection, then the total time until you can
> rebuild your package is very long (for example, I needed to wait for
> all the ocamlnet, equeue, pcre, etc. libs to transition).
> 
> Perhaps we should auto-build everything into experimental first, so
> developers would have a consistent, new environment to build against.
> Anything that fails to rebuild could cause a high-priority bug email
> to the maintainer.  Then after Zack's suggested one week period, the
> successfully built packages (or manually uploaded ones) could go to

You know though that what you propose is what happens for unstable, and that
after 10 days, non-buggyness and consistent build on all RC arches, the
packages moves to testing.

This sounds suspiciously like what you propose, don't you think so ? 

Friendly,

Sven Luther



Reply to: