[PATCH] avoid false positives for addremove -s, mark II

[Erling Ellingsen]

Polished version of the previous patch.

> This looks mostly ok, but gmail has the annoying tendency to mangle
> inline patches.  Can you send it again as an attachment?

Really? Neither 'patch' nor 'hg import' complained at my end. Well, attached.

> Put the import at the top of the file, with the other imports

I wanted to avoid loading an extra file for the 1-ε cases where it's
not used. However, I see that bdiff is included by mdiff anyway, so it
didn't help. Moved.

> This assert shouldn't be needed.  Well, asserts obviously should never
> trigger, but if we really want that check, there are probably better
> places for it.

I put the assert there as I expected the return value of bdiff to
change in the near future (given that it is returning uninitialized
bits of memory right now). Removed.

-- 
 Erling
-------------- next part --------------
A non-text attachment was scrubbed...
Name: similar.patch
Type: application/octet-stream
Size: 4033 bytes
Desc: not available
Url : http://www.selenic.com/pipermail/mercurial-devel/attachments/20070218/1c863482/similar.obj