[Monotone-devel] Re: change_set.cc:526: invariant 'I(j != p2.end())' violated

[Peter Simons]

Are there any insights already what might be causing this
problem? I did a whole bunch of 'renames' in that branch;
that has always meant bad luck in Monotone. :-(

Anyway, now I can't even merge two heads in the branch any
longer:

  $ monotone merge
  monotone: merging with revision 1 / 2
  monotone: [source] 0bc6b6b7067fb91ce933693873d4f4ac58e10dc1
  monotone: [source] 3a96a5b59421e4a05abb61eeb968f0c6ab14a6ab
  monotone: common ancestor 493c28759184d7b0de4944607d8b1ee8440ee063 found
  monotone: trying 3-way merge
  monotone: discarding debug log
  monotone: fatal: std::logic_error: change_set.cc:526:         \
        invariant 'I(j != p2.end())' violated

What is curious about this failed merge is that both
versions are _identical_. I did a "diff -cbr head1 head2" to
verify it. Or had to, actually, because:

  $ monotone diff --revision=0bc --revision=3a9
  monotone: expanded selector '0bc' -> 'i:0bc'
  monotone: expanding selection '0bc'
  monotone: expanded to '0bc6b6b7067fb91ce933693873d4f4ac58e10dc1'
  monotone: expanded selector '3a9' -> 'i:3a9'
  monotone: expanding selection '3a9'
  monotone: expanded to '3a96a5b59421e4a05abb61eeb968f0c6ab14a6ab'
  monotone: discarding debug log
  monotone: fatal: std::logic_error: change_set.cc:526:         \
        invariant 'I(j != p2.end())' violated

Duh!

If it helps, I could make the database file that causes this
available to the developers. I'm also happy to provide any
other information you might need.

Peter