Re: [PATCH v3 0/6] Various fixes for v2.45.1 and friends

From: Junio C Hamano <gitster@pobox.com>
Date: May 21, 2024, 05:33 UTC
Message-ID: <xmqqikz7d90g.fsf@gitster.g>
In-Reply-To: <xmqqseycca0z.fsf@gitster.g>

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

Show 13 quoted lines

> In any case, I just compared the result of applying these patches to
> v2.39.4 with the result of reverting the following out of v2.39.4:
>
>     584de0b4 (Add a helper function to compare file contents, 2024-03-30)
>     8db1e874 (clone: prevent hooks from running during a clone, 2024-03-28)
>     20f3588e (core.hooksPath: add some protection while cloning, 2024-03-30)
>
> and the differences was exactly as I expected.  A Makefile fix and a
> new test added to t1350 are the extra in the series, but otherwise
> the patches are essentially reversion of these three steps.  Very
> nicely done.
>
> Thanks for a quick turnaround.  Will take further look.

I completed merge-up exercise and compared the result with your "tentative" cascade from maint-2.39 to maint-2.45 tracks.

The differences came from pointed cherry-picks (like 'ci: avoid bare "gcc" for osx-gcc job') looked minimal and sensible. I wonder what the best way to do a public review of this kind of history, though.

$ git log --oneline --graph maint-2.45..dscho/tentative/maint-2.45
* aeddcb0275 Git 2.45.2
* 65f0d62523 Sync with 2.44.2
* 9953011fcd Git 2.44.2
* f78818b645 Sync with 2.43.5
* 0aeca2f80b Git 2.43.5
* 0cc3782b1a Sync with 2.42.3
* 33efa2ad1a Git 2.42.3
* 30195eb2b6 Sync with 2.41.2
* 5215e4e368 Git 2.41.2
* 9d6788fd73 Sync with 2.40.3
* 4bf5d57da6 Git 2.40.3
* 9f7a956be5 Sync with 2.39.5
* b9a96c4e5d Git 2.39.5

All of the above (and the one below) are merging up, resolving conflicts, and updating release notes and GIT-VERSION-GEN.

*   b674c6f66c Merge branch 'js/fix-v2.39.4-regressions' into maint-2.39
|\  
| * 5c576e889d Revert "Add a helper function to compare file contents"
| * 0044a35567 clone: drop the protections where hooks aren't run
| * cd14042b06 tests: verify that `clone -c core.hooksPath=/dev/null` works again
| * 57db89a149 Revert "core.hooksPath: add some protection while cloning"
| * 961dfc35f4 init: use the correct path of the templates directory again
| * d4a003bf2c hook: plug a new memory leak

The above 6 patches all appeared on the list in this "v3" thread.

* 883ca51e0a Merge branch 'jk/ci-macos-gcc13-fix' into 'maint-2.39'

This is a merge of the following three patches to maint-2.39

* d4543be3f2 ci: stop installing "gcc-13" for osx-gcc
* 2aef8020d2 ci: avoid bare "gcc" for osx-gcc job
* f3e5bdfebc ci: drop mention of BREW_INSTALL_PACKAGES variable

These three patches were taken from jk/ci-macos-gcc13-fix that was forked from v2.45.0 and rebased them on top of v2.39.4. The bottom one seems to have been adjusted for the older contexts, which during subsequent merging-up has been adjusted back again for the more recent contexts (e.g., we used to use $HOME/bin but use $P4_PATH for Perforce these days, and such differences in the base version appear in the context for "BREW_INSTALL_PACKAGES" change).

So, in short, I didn't see anything unexpected to see in these branches. The "ci" fixes were already reviewed elsewhere (even though there are slight deviations), so if people are OK with the 6 patches in this thread, I would say we are good to go.

Thanks.