Paul Burba wrote:
> > That (and your r1145653 edit) still looks the wrong way around.
> > Because, unless I'm misreading the double-negatives, this function
> > supposedly returns a set of mergeinfo that refers to *non-existent*
> > path-revs.
> Hi Julian,
> Ugh, you're correct, I had it completely backwards.
> I suspect part of the reason this is so confounding is that
> get_invalid_inherited_mergeinfo() answers such an odd question, i.e.
> "What non-existent merge sources does a path inherit?". The sole
> caller has to remove the result from the target's inherited mergeinfo
> to come up with something meaningful.
> In r1148436 I replaced get_invalid_inherited_mergeinfo() with
> remove_non_existent_inherited_mergeinfo(), which instead asks "Remove
> non-existent merge sources from the input mergeinfo". That, to me
> anyway, makes a lot more sense. I also reworked all the comments
> (again). While nothing is radically different with the code,
> hopefully this will save some pain for the next person through this.
Received on 2011-07-19 18:48:04 CEST