Re: [PATCH] experimental: default to fetch.writeCommitGraph=false

From: Taylor Blau <me@ttaylorr.com>
Date: Jul 7, 2020, 16:53 UTC
Message-ID: <20200707165342.GB36941@syl.lan>
In-Reply-To: <xmqqeepn5kxz.fsf@gitster.c.googlers.com>

On Tue, Jul 07, 2020 at 09:50:00AM -0700, Junio C Hamano wrote:

Show 16 quoted lines

> Taylor Blau <me@ttaylorr.com> writes:
>
> >> I wonder if we perhaps wnat to add to the documentation for
> >> writeCommitGraph configuration that its use is currently not
> >> recommended in a shallow clone or something (I know it is not
> >> a problem just to use it with shallow but the breakage needs
> >> to involve unshallowing, but by definition those who do not
> >> use shallow would not hit the unshallowing bug, so...).
> >
> > I think this is a good direction if you don't want to take the patch I
> > sent in [1] for v2.28.0. If you do, though, I don't think that this
> > would be necessary.
>
> Good timing.  I didn't know a "fix" was already being worked on ([1]
> is the patch from this morning, right?  I haven't seen it except for
> its subject).

[1] is the fix. Jonathan wrote it a month or so ago, I just added a test on top. (Independently, I tested it with the reproduction in the original bug report, and it worked properly).

> We could obviously do both excluding it from the usual experimental
> set and applying your fix, so that those who are really curious can
> help us make sure your fix would be all that is needed.  Let's see
> what Jonathan says...

Either of those sound good to me.

> Thanks.

Thanks, Taylor