[MERGE][1.0] New annotate merge that performs well on packs

John Arbash Meinel john at arbash-meinel.com
Mon Dec 3 18:26:19 GMT 2007


John Arbash Meinel has voted comment.
Status is now: Semi-approved
Comment:
So are you saying that even though knits have the annotations already 
cached, it just ignores them?

Certainly things like this:
+        a_ancestry = set(vf.get_ancestry(a_rev))
+        b_ancestry = set(vf.get_ancestry(b_rev))
+        self.uncommon = a_ancestry.symmetric_difference(b_ancestry)

Are ones we want to avoid in the future, but until our Graph support is 
better about giving us the "symmetric_difference" directly, this is 
probably fine for now.


...

You have a lot of calls to "ancestry = set(vf.get_ancestry(revision_id))

Which means that you don't care about the topological order of the 
ancestry. But you *don't* use "get_ancestry(rev_id, topo_sorted=False)"

Otherwise it seems to me that you addressed Robert's concerns. But I'll 
let him give the final decision. (so BB:tweak from me, but pending 
Robert's comments)

For details, see: 
http://bundlebuggy.aaronbentley.com/request/%3C475353A4.4060601%40utoronto.ca%3E



More information about the bazaar mailing list