Re: [PATCH/RFC] add --authorship-order flag to git log / rev-list

From: Elliott Cable <me@ell.io>
Date: Jun 6, 2013, 18:06 UTC
Message-ID: <CAPZ477O2mRCi3gUE+Qoa8Vig2Z2Q5cUzRS0+KEZK3zufmOceig@mail.gmail.com>
In-Reply-To: <7vmwr57lo1.fsf@alter.siamese.dyndns.org>

On Tue, Jun 4, 2013 at 2:53 PM, Junio C Hamano <gitster@pobox.com> wrote:

Show 9 quoted lines

> After reading the subject alone, my reaction was "is this sorting
> commits by the name of the author"?
>
> That is one of the expected natural reactions when people hear about
> this option, which is not what you want.
>
> Perhaps naming it --authordate-order (or enhance the command line
> parsing to allow --date-order=author|committer) would give us a
> better UI.

The same comment was raised by someone in IRC when I submitted an RFC on this. The conclusion we'd arrived at, IIRC, was that the only remotely-not-ugly solutions were either --authorship-order or --author-date-order.

I really like the idea of [--date-order[=author|committer]], but that's getting beyond my knowledge of the code-base. Perhaps I should just implement the changes to the implementation in *my* revision of the patch, and leave it up to a future patcher with the requisite knowledge of the argumentation features to throw in the changes to that flag quickly? Either that, or implement it as --author-date-order right *now*, and change it later before it hits Master (so we don't end up with a no-longer-supported feature?)

(It'd take me many hours to track down the details of how git's codebase goes around doing that, and then attempting to replicate it, whereas someone familiar could probably do it in fifteen minutes, hence the thought-process. Commentary welcome.)