Re: RFE: "git bisect reverse"

From: Ealdwulf Wuffinga <ealdwulf@googlemail.com>
Date: May 31, 2009, 22:41 UTC
Message-ID: <efe2b6d70905311541y2576ef0enc66d619672d4ef34@mail.gmail.com>
In-Reply-To: <4A1F628F.9010407@vilain.net>

On Fri, May 29, 2009 at 5:20 AM, Sam Vilain <sam@vilain.net> wrote:

>
> I mean, can you port your algorithm from python to the git-bisect C
> source and therefore make something of benefit to existing 'git bisect'
> users ?

I haven't really looked at the git bisect code yet, but the algorithms are so different that it would likely make the most sense to just add my algorithm in parallel, rather than trying to combine them. (Although my algorithm can handle the deterministic case as well, you'd want to keep the original one too, because mine starts to get a bit sluggish deciding which commit to test for larger commit ranges. ) It could still be driven by git-bisect script, though.

Converting it to C wouldn't be a trivial task, though, so I may not get around to it. I'll probably write a more git-like frontend, and maybe submit a patch to add it to the contrib directory, to get a better idea of whether there are actually any users for this feature, before deciding whether to do the conversion.

Ealdwulf