Re: Intent to Backport of cryptokit

[Romain Beauxis <toots@rastageeks.org>]

Le Monday 14 July 2008 11:48:04 Stefano Zacchiroli, vous avez écrit :
> On Wed, Jul 09, 2008 at 03:16:20PM +0200, Romain Beauxis wrote:
> > By the way, I would propose that we don't backport ocaml unless there is
> > a major reason for doing it. This would really ease all other backports,
> > and release the need to backport *all* stable ocaml modules against the
> > backported ocaml.
> >
> > What does the team think of this ?
>
> Written as this it sounds like an undeniable desirable property :-)
> I mean: how do you define a "major reason"? The only reason I see, ever,
> for a backport of ocaml is a new upstream release. Our upstream rarely
> does releases for silly reasons, and releases tend to stay stable for
> several months.
>
> Is a new upstream release a major reason in your opinion?

If we backport each upstream release, then there's no point in putting on a 
backport policy like I proposed...

A constructive argument could be established by discussing the changelog of 
the upstream releases. But even the memory bug on amd64 doesn't seem 
sufficient to me. 


Romain