From: Jeff King <peff@peff.net>
Date: Tue, 29 Mar 2011 20:53:08 GMT
Subject: Re: [PATCH] revision.c: introduce --notes-ref= to use one notes ref only
Message-ID: <20110329205307.GA30959@sigill.intra.peff.net>
In-Reply-To: <7vwrjhrd53.fsf@alter.siamese.dyndns.org>

On Tue, Mar 29, 2011 at 11:32:24AM -0700, Junio C Hamano wrote:

> Jeff King <peff@peff.net> writes:
> 
> > Speaking of which, it is kind of weird that --show-notes is negated by
> > --no-notes. So maybe it makes sense to introduce "--notes[=<ref>]" to do
> > what I wrote above, and deprecate --show-notes.
> 
> I think that is sensible.
> 
> I personally think that "notes" are way too premature to be used seriously
> by normal people yet, and if we want to fix UI and semantics warts (it is
> understandable if we had plenty of them, simply because we didn't know
> enough about possible use cases during the period we prototyped the notes
> feature), the time to do so is now.

It is tempting to kill off --show-notes and --standard-notes, as it
would clean up this code a bit. But even though they are probably not
being seriously used by normal people, they have been in several
released versions.

Here's the series I ended up with. Getting the refactoring just right
turned out to be non-trivial, but between several attempts and some
tests, I think the end result is correct. Hopefully the breakdown of the
changes into small patches helps make it easy to review.

  [1/6]: notes: make expand_notes_ref globally accessible
  [2/6]: revision.c: refactor notes ref expansion
  [3/6]: notes: refactor display notes extra refs field
  [4/6]: notes: refactor display notes default handling
  [5/6]: revision.c: support --notes command-line option
  [6/6]: revision.c: make --no-notes reset --notes list

-Peff