On Tue, Jun 10, 2014 at 11:03:09PM +0200 I heard the voice of
Rhialto, and lo! it spake thus:
> 
> I'd say to do it sooner rather than later. I expect I can handle the
> fixup that is needed for my branch, and as I'm adding stuff the
> changes in the files other than ewmh.[ch] would only get more, even
> if it's only some #ifdef protected call to a new function here and
> there.

In some additional trials, the following dance seems to get correct
results without a lot of manual fiddlery.


1) Reindent trunk (say this moves from rev 555 to 556, just to make up
   numbers).

2) Merge trunk r555 (pre-reindent) into outstanding branch, dealing
   with whatever conflicts may come up.  So far this is pretty normal.
   Be sure this gets committed before moving on to

3) Reindent branch.

4) Merge trunk r556 (post-reindent; pre any other later changes made).
   This will throw up a pile of conflicts because most of the lines in
   trunk have changed, which means any of your changes end up
   conflicting.  BUT, since you just handled all the real diffs in the
   previous merge, the only changes in trunk you're dealing with here
   are the result of the reindent.  And since you did the same
   reindent locally, that means the actual files shouldn't change at
   all as a result of this merge.

   So, you tell bzr to always choose your local changes in conflicts,
   by running "bzr resolve --take-this".  Then, running status should
   show no file changes, just the pending merge:

--------
% bzr stat
pending merge tips: (use -v to see all merge revisions)
  Matthew Fuller 2014-06-11 Restyle.
--------

   and then commiting that and moving forward.  Possibly then merging
   later trunk changes, if 557+ have come around yet.


This works right for me in a little testing.  I _think_, if my mental
algebra is reliable, it should guarantee D'ingTRT, as long as trunk
555->556 is specifically and only the reindent, your (3) uses the same
reindent params, and you merge just those revs in that order.

'd be good to have somebody double-check my reasoning.


-- 
Matthew Fuller     (MF4839)   |  [email protected]
Systems/Network Administrator |  http://www.over-yonder.net/~fullermd/
           On the Internet, nobody can hear you scream.

Reply via email to