Re: [PATCH 0/2] describe and diff: implement --no-optional-locks

From: Jeff King <peff@peff.net>
Date: Mar 10, 2025, 16:08 UTC
Message-ID: <20250310160837.GA26308@coredump.intra.peff.net>
In-Reply-To: <xmqqwmcx9io3.fsf@gitster.g>

On Mon, Mar 10, 2025 at 05:25:32AM -0700, Junio C Hamano wrote:

Show 14 quoted lines

> Jeff King <peff@peff.net> writes:
> 
> > .... What
> > git-status does with --no-optional-locks is to update the index
> > internally for its _own_ use (giving it the correct results), but not to
> > lock nor write out the resulting index (to avoid conflicting with other
> > running programs). So it's pessimal (losing the opportunity to share
> > what it learned) but prevents lock contention.
> 
> Yup, that sounds somewhat sensible.  I also have to wonder, other
> than commands that are clearly about changing the repository state
> like "add", the inspection commands like diff and status should
> always opportunistically write the index back, without even being
> asked?

Yeah, certainly it is a potential source of confusion to have a conceptually read-only operation take locks or modify the repo state.

But I'm not sure we have a sense of how valuable that optimization is in practice. After touching some files, every git-status, git-diff, etc would end up re-hashing those files instead of using the stat cache. But maybe that is lost in the noise of reading the files to actually do diffs, etc? I dunno. I expect it is more important for status, which probably does not need to read the whole file contents in most cases (and which may be run a lot from the user's prompt, etc).

It seems like a big and possibly risky departure from what we've done for so many years. I'm inclined not to rock the boat too much. ;)

-Peff