Re: [RFC 3/3] refspec: add support for negative refspecs

[Jeff King <peff@peff.net>]

On Thu, Aug 20, 2020 at 04:59:53PM -0700, Jacob Keller wrote:

> > The relevant commit is 2467a4fa03 (Remove duplicate ref matches in
> > fetch, 2007-10-08), I think. We may end up with multiple refspecs
> > requesting a particular ref. E.g.:
> >
> >   git fetch origin refs/heads/master refs/heads/*
> >
> > I don't think the order should matter. If we apply negative refspecs
> > first, then we'd either remove both copies or leave both untouched (and
> > if the latter, then de-dup to a single). If we apply negative refspecs
> > after de-duping, then we'd either remove the single or leave it in
> > place. But the result is the same either way.
> 
> I'm not sure this is quite true in the case where destinations are
> supplied. Suppose this case:

Oh, you're right. I was too focused on the de-duping of identical refs,
but this is also handling colliding destinations.

> git fetch refs/heads/*:refs/remotes/origin/*
> refs/other/mybranch:refs/remotes/origin/mybranch
> 
> This would ofcourse error out due to de-duping where we determine that
> both would fetch to the same place.. however if you also added a
> negative refspec:
> 
> git fetch refs/heads/*:refs/remotes/origin/*
> refs/other/mybranch:refs/remotes/origin/mybranch ^refs/heads/mybranch
> 
> then shouldn't this work? meaning we should de-dupe only after we
> apply negative refspecs in this case?

Yes, I'd agree we should be applying the negative refspecs first, and
then de-duping / looking for collisions. Which I think is what the patch
is doing currently.

-Peff