Re: What's cooking in git.git (Feb 2022, #07; Fri, 25)

From: Tao Klerks <tao@klerks.biz>
Date: Mar 1, 2022, 19:25 UTC
Message-ID: <CAPMMpoiexaYigvko1cD=qZAbdrso06Ck+T=j4Ai68qSaKkXVrA@mail.gmail.com>
In-Reply-To: <xmqqfso1r1hp.fsf@gitster.g>

Hmm, does this work, or still too long?

   The performance of the "untracked cache" feature is improved when
   "status.showUntrackedFiles" is configured to "all", unless
   "--untracked-files=normal" is also specified, in which case it is
   degraded. This change targets -uall users, who can now configure
   to benefit from untracked cache.

Side-note: your reaction wrt agreement on "thought not to matter"
makes me think maybe making this behavior fully optional, dependent on
a *new* config value, would be better? I've assumed adding more config
would be frowned upon, but maybe the reduction in risk to a specific
possibly-existing usage pattern makes it worthwhile?

I'm thinking a config value like "status.optimizeUntrackedCacheFor" with values "normal" and "auto", where the current effective default is "normal", and users who need "-uall" to work can set "auto" (for the behavior as I proposed/implemented it so far). Would the increase in complexity be warranted by the reduction in risk and corresponding reduction in need for explanation? The resulting topic summary could be:

   New option to support/optimize untracked cache for
   "status.showUntrackedFiles=all".

On Tue, Mar 1, 2022 at 7:06 PM Junio C Hamano <gitster@pobox.com> wrote:

Show 35 quoted lines

>
> Tao Klerks <tao@klerks.biz> writes:
>
> > On Mon, Feb 28, 2022 at 2:47 PM Ævar Arnfjörð Bjarmason
> > <avarab@gmail.com> wrote:
> >>
> >>
> >>
> >> Perhaps something like this:
> >>
> >>     The performance of the "untracked cache" feature has been improved in
> >>     common cases where "--untracked-files=<mode>" and
> >>     "status.showUntrackedFiles" were combined. This change benefits Windows
> >>     users using it in conjuction with the "fsmonitor feature in particular.
> >>
> >> Perhaps adding:
> >>
> >>     There's an obscure case where the performance is now worse, but it's
> >>     thought not to matter.
> >>
> >
> > I really like it, thx! I would change "combined" to "aligned", and little else:
> >
> >    The performance of the "untracked cache" feature has been improved in
> >    common cases where "--untracked-files=<mode>" and
> >    "status.showUntrackedFiles" were aligned. This change benefits Windows
> >    users using it in conjunction with the "fsmonitor" feature in particular.
> >    There's an obscure case where the performance is now worse, but it's
> >    thought not to matter.
>
> Overly long for a topic summary, with value judgements that I'd
> prefer to omit.  What (you think) is "common" does not help us make
> a stronger case.  I've never said I agree "it's thought not to
> matter" at all.  Just stick to the known facts.
>