Re: [PATCH] log: add %S option (like --source) to log --format

From: Issac Trotts <issac.trotts@gmail.com>
To: Jeff King <peff@peff.net>
Cc: Junio C Hamano <gitster@pobox.com>,
	git@vger.kernel.org, Noemi Mercado <noemi@sourcegraph.com>
Subject: Re: [PATCH] log: add %S option (like --source) to log --format
Date: Wed, 19 Dec 2018 00:07:35 -0800	[thread overview]
Message-ID: <CANdyxMzYMc2DGC8JUE7N6Ry7iOLH09SwR5OFQ64c1vtbO388FA@mail.gmail.com> (raw)
In-Reply-To: <CANdyxMzOpRRu+B0cq5cbrmYsy47cod8N4mWrGtfOgRRgKVh+xA@mail.gmail.com>

Travis showed one failure. I'm not sure if it's a flake.
https://travis-ci.org/ijt/git/builds/469843833. Rerunning it.

On Tue, Dec 18, 2018 at 7:47 PM Issac Trotts <issac.trotts@gmail.com> wrote:
>
> On Tue, Dec 18, 2018 at 9:14 AM Issac Trotts <issac.trotts@gmail.com> wrote:
> >
> > Hi Peff, thanks for the feedback. I tried a variant of the command you
> > showed and it yielded a seg fault:
> > ```
> > [ issactrotts ~/git ] ./git diff-tree -s --pretty=tformat:'%S %H %s' HEAD
> > Segmentation fault: 11
> > ```
> > I'll see if I can track it down this evening.
>
> Okay, found it. My solution is to make %S not do anything for commands
> other than `log` since `log` is the only one that has --source defined
> as far as I can tell.
>
> I'm waiting for Travis to run and will post an updated patch if
> everything looks good there.
>
> > Issac
> >
> > On Mon, Dec 17, 2018 at 7:59 AM Jeff King <peff@peff.net> wrote:
> > >
> > > On Sun, Dec 16, 2018 at 10:25:14PM -0800, Issac Trotts wrote:
> > >
> > > > Make it possible to write for example
> > > >
> > > >         git log --format="%H,%S"
> > > >
> > > > where the %S at the end is a new placeholder that prints out the ref
> > > > (tag/branch) for each commit.
> > >
> > > Seems like a reasonable thing to want.
> > >
> > > One curious thing about "--source" is that it requires cooperation from
> > > the actual traversal. So if you did:
> > >
> > >   git rev-list | git diff-tree --format="%H %S"
> > >
> > > we don't have the %S information in the latter command. I think that's
> > > probably acceptable as long as it does something sane when we don't have
> > > that information (e.g., replace it with an empty string). It might also
> > > be worth calling out in the documentation.
> > >
> > > -Peff