Re: feature suggestion: optimize common parts for checkout --conflict=diff3

From: Jeff King <peff@peff.net>
Date: Mar 6, 2013, 20:54 UTC
Message-ID: <20130306205400.GA29604@sigill.intra.peff.net>
In-Reply-To: <7vvc94p8hb.fsf@alter.siamese.dyndns.org>

On Wed, Mar 06, 2013 at 12:40:48PM -0800, Junio C Hamano wrote:

Show 10 quoted lines

> Jeff King <peff@peff.net> writes:
> 
> > then it will produce the output that Uwe expects. While it can be
> > misleading,...
> 
> Misleading is one thing but in this case isn't it outright wrong?
> 
> If you remove <<< ours ||| portion from the combined diff output,
> I would expect that the hunk will apply to the base, but that is no
> longer true, no?

It shifts the concept of what is the "base" and what is the "conflict". In Uwe's example, no, it would not apply to the single-line file that is the true 3-way base. But it would apply to the content that is outside of the hunk marker; we have changed the concept of what is in the base and what is in the conflict by shrinking the conflict to its smallest size. The same is true of the conflict markers produced in the non-diff3 case. It is a property of XDL_MERGE_ZEALOUS, not of the conflict style.

If your argument is "diff3 means something different than regular conflict markers; it should have the property of being machine-convertible into a patch, but regular markers do not", I'm not sure I agree. It may be used that way, but I think it is mostly used in git to give the reader more context when making a resolution. And anyway, I think the proposed change would not be to change diff3, but to introduce a new diff3-like format that also shrinks the hunk size, so it would not hurt existing users of diff3.

-Peff