bzr.dev has broken 'bzr log --short' (it shows merges)

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I realize this may be somewhat intentional, but the output of "bzr log
- --short" just changed in bzr.dev.

Specifically, it is now showing merged revisions. (Which means it also
computes dotted revnos, and takes 2.6s instead of 0.4s for 'bzr log
- --short --forward -r-10..-1').

I don't see a new flag that allows me to say "I really don't want merged
revisions", so I have the feeling something was accidentally broken in
Ian's recent revisions, which caused "--short" to no-longer indicate
that we didn't want to show merged revisions.

I'm not sure how this wasn't caught, though I wouldn't be surprised if
the test suite wasn't explicitly testing that with log --short isn't
showing merged revs.

I found the bug, which is that we have these lines:

    include_merges = generate_merge_revisions or specific_fileid
    view_revs_iter = get_view_revisions(mainline_revs, rev_nos, branch,
                          direction, include_merges=include_merges)

And in get_view_revisions we have:
        if include_merges is False:


So for "bzr log --short -r -10..-1" we actually have

generate_merge_revisions = False
and
specific_fileid = None

Which means that include_merges is actually "None" and not False. (I'm
guessing Ian didn't entirely realize the pythonism:

  (a or b) is b
  (a or b) is not False

Or

  (False or None) is None

the "or" operator actually returns the second object, not a boolean value.

Anyway, I prefer:

  if not include_merges:

Anyway, rather than "is False".


Because this is broken for me "right now", I'm going to submit this patch:

=== modified file 'bzrlib/log.py'
- --- bzrlib/log.py       2009-01-15 14:05:13 +0000
+++ bzrlib/log.py       2009-01-16 22:14:55 +0000
@@ -620,7 +620,7 @@
     :return: an iterator of (revision_id, revno, merge_depth)
     (if there is no revno for a revision, None is supplied)
     """
- -    if include_merges is False:
+    if not include_merges:
         revision_ids = mainline_revs[1:]
         if direction == 'reverse':
             revision_ids.reverse()

And then hopefully I'll get a proper patch with a test that we can review.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAklxBxkACgkQJdeBCYSNAAPhNwCff64b4rZVgC1U7TOoSKE72QsO
cGwAoM7lBRgZekSqGHLTC/Mg68qIcAel
=YUGd
-----END PGP SIGNATURE-----