Re: git status always modifies index?

From: Jonathan Nieder <jrnieder@gmail.com>
To: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Cc: Jeff King <peff@peff.net>, Nathan Neulinger <nneul@neulinger.org>,
	Santiago Torres <santiago@nyu.edu>,
	git@vger.kernel.org
Subject: Re: git status always modifies index?
Date: Mon, 27 Nov 2017 12:49:09 -0800	[thread overview]
Message-ID: <20171127204909.GA27469@aiede.mtv.corp.google.com> (raw)
In-Reply-To: <alpine.DEB.2.21.1.1711272142120.6482@virtualbox>

Hi,

Johannes Schindelin wrote:
> On Mon, 27 Nov 2017, Jeff King wrote:

>> [...] IMHO it argues for GfW trying to land patches upstream first, and
>> then having them trickle in as you merge upstream releases.
>
> You know that I tried that, and you know why I do not do that anymore: it
> simply takes too long, and the review on the list focuses on things I
> cannot focus on as much, I need to make sure that the patches *work*
> first, whereas the patch review on the Git mailing list tends to ensure
> that they have the proper form first.
>
> I upstream patches when I have time.

You have been developing in the open, so no complaints from me, just a
second point of reference:

For Google's internal use we sometimes have needed a patch faster than
upstream can review it.  Our approach in those cases has been to send
a patch to the mailing list and then apply it internally immediately.
If upstream is stalled for months on review, so be it --- we already
have the patch.  But this tends to help ensure that we are moving in
the same direction.

That said, I don't think that was the main issue with
--no-optional-locks.  I'll comment more on that in another subthread.

Thanks,
Jonathan