Re: reference-transaction regression in 2.36.0-rc1

From: Patrick Steinhardt <ps@pks.im>
To: Michael Heemskerk <mheemskerk@atlassian.com>
Cc: "Junio C Hamano" <gitster@pobox.com>,
	"Ævar Arnfjörð Bjarmason" <avarab@gmail.com>,
	"Bryan Turner" <bturner@atlassian.com>,
	"Git Users" <git@vger.kernel.org>
Subject: Re: reference-transaction regression in 2.36.0-rc1
Date: Wed, 27 Apr 2022 13:05:35 +0200	[thread overview]
Message-ID: <YmkjfyB3gIQ2t45V@ncase> (raw)
In-Reply-To: <CAJDSCnOUQm-doY-xTobPk64oeiSsTd+vSFzsb_cz9BLORAxXGA@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 1311 bytes --]

Thanks a lot to all of you for handling this regression while I was out
of office!

On Tue, Apr 19, 2022 at 11:54:19AM +0200, Michael Heemskerk wrote:
> Apologies for the late reply. Bryan had his last day at Atlassian last
> week, so I'll take over from him on this topic.
> 
> Thanks for asking this question, which is lot more urgent and I
> > should have asked last night before I went to bed.  Yes, Bryan
> > already said that the revert in 'seen' the previous day made their
> > system happier, but it certainly helps to know the reverts were OK
> > at the tip of 'master' without all the other random topics.
> >
> 
> I have run all our tests on the tip of master prior to 2.36 being released,
>  and on the released 2.36. Our tests are happy on both, so thanks for
>  reverting those reference-transaction changes for 2.36.
> 
> I'd be happy to provide the fix to files_delete_refs in a patch (including
> some extra tests) on top of Patrick's
> avoid-unnecessary-hook-invocation-with-packed-refs series if and
> when that series is unreverted on 'next'.

That would be great. To be clear, do the fixes you have also fix the
pre-v2.36.0 issues Bryan has mentioned?

Please let me know whether you want to pursue this and whether you need
any further help with it.

Patrick

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]