Re: jn/experimental-opts-into-proto-v2, was Re: What's cooking in git.git (May 2020, #09; Tue, 26)

From: Junio C Hamano <gitster@pobox.com>
Date: May 28, 2020, 14:44 UTC
Message-ID: <xmqq5zcg5db4.fsf@gitster.c.googlers.com>
In-Reply-To: <nycvar.QRO.7.76.6.2005272159570.56@tvgsbejvaqbjf.bet>

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

Show 9 quoted lines

>> No.  I fully accept your reasoning in the proposed log message why a
>> handcrafted query to the config system is done in the location the
>> patch adds a call.
>
> Now, I apologize. I had not reviewed the patch, and only just read it.
>
> I agree that it is a bit unfortunate that it uses such a non-standard way
> that hard-codes "feature.experimental" in a different place than
> repo-settings.c.

You make it sound like it was a choice made by the implementation, but (1) a "non-standard" way may not have to stay non-standard forever (there may be many more experimental features that are not tied to a specific repository in the future), and (2) the patch needs to do it in a way that is not tied to a single repository because it is not at per-repository level decision. As long as we are aware of this limitation caused by the current "experimental" arrangement that is tied to a repository and can work towards extending it to support this new use case in the future, I do not think it is unfortunate at all.

> Had it been a patch to repo-settings.c, I would now have tried to lobby
> for including it into v2.27.0, but as it is, I fully agree with your
> reasoning to just leave it out.

No need to apologize for raising it as an issue---hearing from those with different risk tolerance from time to time is a good way to calibrate my own.

Thanks.