BugDir._in_memory was a stupid idea. Took it back out.
authorW. Trevor King <wking@drexel.edu>
Mon, 27 Jul 2009 00:04:32 +0000 (20:04 -0400)
committerW. Trevor King <wking@drexel.edu>
Mon, 27 Jul 2009 00:04:32 +0000 (20:04 -0400)
commit058dbd6da7c085674680254b68bdba8a6acda117
tree56f474ffdf7ad66feb5135d32fa6af8bf86c2b50
parentfa7dd91cd88377e8c41967ef00a9d31e44cb697b
BugDir._in_memory was a stupid idea.  Took it back out.

It was too confusing having three memory access levels:
  1) syncronized
  2) explicit
  3) memory-only
with .sync_with_disk selecting between 1 and 2/3 and ._in_memory
selecting between 2/3.

Now there are only two:
  1) syncronized
  2) memory-only excepting explicit BugDir.save() calls.

I avoid the problem of non-syncronized loading of on-disk bugs in
simple_bug_dir by restricting .list_uuids() to in-memory bugs when
.sync_with_disk==True.

Beyond that, I shifted the order of the BugDir methods around a bit so
that they were better grouped according to general idea.

Note that the DiskAccessRequired exceptions on filesystem access when
.sync_with_disk==False should be propogated to the Bug and Comment
methods, but I haven't done that yet.
libbe/bugdir.py