Subject: Re: [boost] Fwd: Boost.Python build error in 1.58 RC1, 2 and 3 (OK in Beta 1)
From: Andrey Semashev (andrey.semashev_at_[hidden])
Date: 2015-04-15 11:34:13
On Wednesday 15 April 2015 18:14:35 Peter Dimov wrote:
> Andrey Semashev wrote:
> > I suspect we won't be able to get away with just the superproject
> > branching. If instead of reverting a broken submodule we want to fix it,
> > and the maintainer is unresponsive for some reason, we need either a
> > branch in the submodule or a fork of it. Since the branch can be modified
> > by someone else than a release manager, the fork seems more preferred. The
> > release branch of superproject would have to refer to the fork in this
> > case.
>
> In what way is this an argument against the suggested structure? It supports
> this workflow. I don't think that something like that will ever prove
> necessary, but it's possible.
I didn't say I disagree with your superproject structure. The post you
referenced only described downgrading submodules as a measure of fixing
breakage, I'm just saying we might need to apply patches instead. For that it
is not sufficient to only manage the superproject.