Re: git diff too slow for a file

From: Junio C Hamano <gitster@pobox.com>
Date: Apr 17, 2010, 17:10 UTC
Message-ID: <7vpr1y2eev.fsf@alter.siamese.dyndns.org>
In-Reply-To: <4BC9D928.50909@lsrfire.ath.cx>

René Scharfe <rene.scharfe@lsrfire.ath.cx> writes:

René Scharfe <rene.scharfe@lsrfire.ath.cx> writes:

Show 8 quoted lines

> Am 29.03.2010 03:42, schrieb SungHyun Nam:
>>  ...
>> Well, though the files are ascii file, they includes a random
>> hexa-decimal datas, so that I don't interest the diff result at
>> all.
> ...
> The following patch is not meant for inclusion, but rather to start a
> dicussion.  Is XDF_NEED_MINIMAL a good default to have?

This is a very valid question to ask. The choice of the default was done without any benchmarking nor analysis on performance impact at all.

What we should do next would be to:

 - see how much performance impact we have been getting from more normal
   set of files (say, "git log -p" in the kernel archive) by our use of
   MINIMAL;  I suspect that git.git itself is too small to observe any
   meaningful difference.  We already _know_ that MINIMAL is more
   expensive, so this is not very important, but it would be good to
   know.

 - inspect the difference of the quality of output for not using MINIMAL,
   again for more normal set of files.  We know that the quality does not
   matter for pathological cases like the one in this thread --- the user
   is not even "interested in the diff result at all".

Thanks for starting this.