Re: The criss-cross merge case

From: Benedikt Schmidt <ry102@rz.uni-karlsruhe.de>
Date: Apr 28, 2005, 02:15 UTC
Message-ID: <87d5sf7il2.fsf@rzstud4.rz.uni-karlsruhe.de>
In-Reply-To: <Pine.LNX.4.21.0504272051390.30848-100000@iabervon.org>

Daniel Barkalow <barkalow@iabervon.org> writes:

Show 10 quoted lines

> On Wed, 27 Apr 2005, Tupshin Harper wrote:
>
>> Can you clarify what you mean by darcs' underlying diff not being that
>> great? It seems to function pretty much identically to gnu diff. In what
>> way would you want the underlying diff to be improved?
>
> GNU diff uses an algorithm which is tuned to handle finding the shortest
> diff among a large set of similar-length alternatives while comparing
> files which have a lot of repeated lines. The author of the paper it cites
> is really thinking about diffing DNA sequences or similar things.

AFAIK the paper mentioned in the GNU diff sources [1] is an improvement to an earlier paper by the same author titled "A File Comparison Program" - Miller, Myers - 1985.

Can you be more specific why the algorithm is a bad choice (performance, quality of diff output)?

> It also can't detect content moves, which are a common thing to have, and
> which will be important in the long run, when we're trying to track
> modifications to content which also moved from place to place.

Ok, darcs doesn't handle block moves, so there is no need for an algorithm that supports them (yet). Is there any free SCM that has support for block moves at the moment? It seems like clearcase detects them, but I don't know where it takes advantage of it.

Benedikt

[1] http://citeseer.ist.psu.edu/myers86ond.html