Re: reference-transaction regression in 2.36.0-rc1

[Junio C Hamano <gitster@pobox.com>]

Junio C Hamano <gitster@pobox.com> writes:

> Junio C Hamano <gitster@pobox.com> writes:
>
>> Reverting the merge 991b4d47f0a as a whole is an option.  It might
>> be the safest thing to do, if we do not to want to extend the cycle
>> and add a few more -rc releases before the final.
>
> It turns out that this involves nontrivial amount of work to get
> right, as the bottom commit of Patrick's "fetch --atomic" series
> wants to count how many transaction we make, instead of ensuring
> that the updates to the references are all-or-none, so at least
> 2a0cafd4 (fetch: increase test coverage of fetches, 2022-02-17)
> needs to be reverted as well.  This in turn is made unnecessarily
> more cumbersome as the history in the t/ directory is littered with
> unrelated "clean-up" patches since these topics were merged.

I have a tentative revert of the merge and also the "fetch --atomic"
test that (unnecessarily) counted the number of transactions directly
on top of planned 'master', which merges the planned v2.35.3 into
v2.36-rc2 and pushed it as if it were the 'seen' branch.

The CI job https://github.com/git/git/actions/runs/2164208587 seems
to be doing well so far, so once the dust from releasing v2.30.4,
v2.31.3, v2.32.2, v2.33.3, v2.34.3, and v2.35.3 with Derrick's
hotfix for yesterday's CVE fix, I may merge it down to 'master'
before the final.

Thanks.