But there are still cases where we need to write back even an unmodified
  file. During the open we remove any pending stores from the CML (store
  optimization), and if that happened we have to write back the file even
  if the current open for write didn't change anything.

Probably this removal is not quite right; deferring the pending store
a bit is ok.

consider:

  open for write
  write
  close

  sleep 10

  open for write
  while (long time)
    read some
    sleep
  close

In this case, I'd feel that coda is being broken if the first write
doesn't make it back within the reintegrate timer.  (I'd agree if you
said my program was odd, but that's not the point.)

collapsing two stores actually in the CML seems fine.

-- 
    Greg Troxel <gdt_at_ir.bbn.com>