Re: What's cooking in git.git (Nov 2013, #01; Fri, 1)

[Michael Haggerty <mhagger@alum.mit.edu>]

On 11/02/2013 12:09 AM, Thomas Rast wrote:
> Junio C Hamano <gitster@pobox.com> writes:
>
>> * tr/merge-recursive-index-only (2013-10-28) 3 commits
>>   - merge-recursive: -Xindex-only to leave worktree unchanged
>>   - merge-recursive: internal flag to avoid touching the worktree
>>   - merge-recursive: remove dead conditional in update_stages()
>
> FWIW -- since you dropped the earlier note to the same effect -- I still
> consider this "hold until Michael finds a use for it".

Thomas,

Sorry, I didn't realize that you are waiting on me.  I've been very busy 
lately and just got back from a trip to San Francisco.  The chaos is 
scheduled to continue for at least another week.

The biggest win from being able to run test merges without touching the 
tree would be to move a lot of the time-consuming processing work of 
git-imerge into the background while the user is asked to resolve 
conflicts.  It's kindof awkward, though, because for the user to resolve 
a conflict control has to be passed back to the user.  So the main 
git-imerge process would have to spawn off a worker daemon process that 
would continue working in the background.  But then there has to be some 
kind of communication between the worker thread and the user's 
*subsequent* call to "git imerge continue".

So the bottom line is that it is nontrivial to implement this so I will 
need a bit of concentrated time to dedicate to git-imerge before I will 
be able to give you any feedback.

Michael

-- 
Michael Haggerty
mhagger@alum.mit.edu
http://softwareswirl.blogspot.com/