[Svnmerge] Ignore initialized revisions (was Re: Reflected blocks)

[Dustin J. Mitchell]

On Mon, Jul 23, 2007 at 03:35:28PM +0100, Michael Willmott wrote:
> Now that the testsuite is passing again, Rich's patch (as is) causes a 
> number of failures in the testsuite.  Specifically 1 failure due to 
> changed behaviour, and a number of errors related to return values from 
> analyze_revs.  I've produced an additional patch to fix these failures. 
>   When it comes to reviewing (and committing) these, would you prefer 3 
> separate patches (Rich's original patch, my fixes, and a new testcase), 
> or a single combined patch ?

Speaking for myself, I like patches that are *conceptually*
non-overlapping.  That is, a patch should have a succinct purpose, and
all of the hunks in the patch should visibly address that purpose.

In this case, I think that a single patch would still fit this
definition.

Dustin

-- 
        Dustin J. Mitchell
        Storage Software Engineer, Zmanda, Inc.
        http://www.zmanda.com/