Stefan Sperling wrote:
> On Thu, Nov 27, 2008 at 10:26:15PM +0000, Julian Foad wrote:
>> I have started a patch, attached, to skip obstructions and not allow
>> them to be detected as tree conflicts. This task is issue #3161. My
>> approach is to use a single function that checks the expected node kind
>> and the actual node kind on disk, and raises a flag if they differ.
>>
>> I have made merge_tests 19 pass with this patch. It makes some other
>> tests fail, and I haven't yet investigated them.
>>
>> Review/comments/further work most welcome.
>
> Just a comment:
>
> Looks very nice to me. I do actually wonder why a neat separation
> between obstructed and missing was not present before.
> Maybe because previously, the merge skipped these items in either case?
>
> Also, do we need a similar fix for update/switch?
Yes, we did, until sbutler fixed it. :)
~Neels
Received on 2008-11-27 23:54:00 CET