Re: [BUG] Re: Git 2.35.0-rc0

From: Junio C Hamano <gitster@pobox.com>
Date: Jan 13, 2022, 18:03 UTC
Message-ID: <xmqqv8ynikdm.fsf@gitster.g>
In-Reply-To: <220113.864k67vkea.gmgdl@evledraar.gmail.com>

Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:

> https://lore.kernel.org/git/87wnn62nhp.fsf@evledraar.gmail.com/ this use
> of uncompress2() is just saving a few lines of boilerplate instead of
> using the underlying zlib functions, which every other in-tree user
> uses.

I looked at the source of uncompress2(), and I tend to agree that, we do NOT HAVE TO add dependency on it. You should be able to rewrite the new caller without using it.

But this is a 5-year old API function, that first appeared how many years ago in their public release? In a few years, I would say that we would be laughed at if we avoid it with "it is not available everywhere".

And in the meantime, we ship a copy lifted from upstream, so those with older zlib would be OK with it, too.

That is how I view what we have today. To me, the logical conclusion of the observation is that, we do not have good reason to avoid it.

When we are adding a new piece of code that drives inflate(), we should remember that we now have uncompress2() available in the codebase, and see if we can simplify it by using uncompress2(). The same is true for a case where we refactor existing code that can become simpler with uncompress2().

Thanks.