Re: [PATCH v2 00/10] compat/zlib: allow use of zlib-ng as backend

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

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

> Patrick Steinhardt <ps@pks.im> writes:
>
>> Changes in v2:
>>   - Wire up zlib-ng in our Makefile.
>>   - Exercise zlib-ng via CI by adapting our "linux-musl" job to use
>>     Meson and installing zlib-ng.
>>   - Link to v1: https://lore.kernel.org/r/20250110-b4-pks-compat-drop-uncompress2-v1-0-965d0022a74d@pks.im
>>
>> The series is built on top of fbe8d3079d (Git 2.48, 2025-01-10) with
>> ps/meson-weak-sha1-build at 6a0ee54f9a (meson: provide a summary of
>> configured backends, 2024-12-30) merged into it.
>
> I think you are now also textually depending on the fuzzer thing due
> to touching meson_options.txt and ci/run-build-and-tests.sh with a
> later step.
>
>>  -:  ---------- >  9:  7ae8f413d4 ci: switch linux-musl to use Meson
>>  -:  ---------- > 10:  2dd1b49e4f ci: make "linux-musl" job use zlib-ng
>
> I will see what other things I can find.

Yup.  The patch series for some reason still does not seem to apply
cleanly ([09/10] ci/run-build-and-tests.sh somehow seems to be
troublesome), but it was easy to wiggle it in when the base was
prepared with these two topics merged on top of 'master':

    4610af08e7 ci: make "linux-musl" job use zlib-ng
    b2ddd0b33e ci: switch linux-musl to use Meson
    5118183ef4 compat/zlib: allow use of zlib-ng as backend
    08bf6b2062 git-zlib: cast away potential constness of `next_in` pointer
    ebf98412e3 compat/zlib: provide stubs for `deflateSetHeader()`
    29829e5714 compat/zlib: provide `deflateBound()` shim centrally
    8f19b26bbe git-compat-util: move include of "compat/zlib.h" into "git-zlib.h"
    8aab230253 compat: introduce new "zlib.h" header
    1ce001beaa git-compat-util: drop `z_const` define
    b9d4bd5467 compat: drop `uncompress2()` compatibility shim
    db620fad21 Merge branch 'ps/build-meson-fixes' into ps/zlib-ng
    64156589d9 Merge branch 'ps/meson-weak-sha1-build' into ps/zlib-ng

I think the reason is because the other topic that touches the fuzz
thing we see in the context of [09/10] is not ps/build-meson-fixes
but something else that is before "--fatal-meson-warnings" was
added.

One request.  You seem to have started using --full-index when
generating the patches.  It is extremely annoying when a patch needs
to be mucked with an editor to inspect why it does not apply and to
tweak it to make it apply.  40-hex does not help at all if the base
commit is not conveyed correctly, as the recipient will not have the
necessary blob objects _anyway_.  And 40-hex is unnecessarily long
in order to protect the recipient who uses "--3way" from using a
wrong blob in a fake ancestor tree.  Please stop.

Thanks.