From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Fri, 06 Jun 2008 14:04:01 GMT
Subject: Re: [PATCH v2 1/2] Allow git-apply to ignore the hunk headers
Message-ID: <alpine.DEB.1.00.0806061502030.1783@racer>
In-Reply-To: <4848E105.7050405@gnu.org>

Hi,

On Fri, 6 Jun 2008, Paolo Bonzini wrote:

> > @@ -0,0 +0,0 @@
> > 	default:
> > -		return -1;
> > +		return len != 4 && memcmp(line - len, "-- \n", len);
> >   }
> 
> You're never returning -1 here, right?

You are a clever guy!  I really do not return -1 here.  But then, the 
return value is only checked for non-zeroness.  As is obvious from the 
part you did not quote.

> > However, this will not work if anybody has a signature starting with 
> > "@@ ", "+", " ", "-" or "diff "...
> 
> I think that the main worry is the patches made with git-format-patch, 
> and those are not problematic.

Actually, this change was done in v3 on _explicit_ request from Junio who 
wants to be able to use the patch for git-am, where we cannot rely on 
format-patch.

So yes, they _are_ problematic.

Thanks,
Dscho