Jack,
205, 226: How come for some cases (e.g. line 189 for partition resize)
self._children = ShadowPhysical(self)
but other times this is not necessary? Both cases are deleting the
existing partition and then calling self.parent.add_partition().
Seems like it is related to whether or not
self.parent.insert_children() is called.
That's a bug in Partition.resize(). The delete() call will take care of
deleting the Partition object and all children. When
parent.add_partition() is called, a new Partition object will be created
with _children set to a ShadowPhysical() list. I'll remove that call.
But if this is the case, what happens when change_type is called
(which also delete()s and add_partition()s), followed by a call to
add_slice() which calls insert_children() but does not call
ShadowPhysical() first?
See above. Taking the assignment out of both resize() calls (Partition
and Slice).
241-251 and 270: Assumes there will be no errant or overlapping
partitions (i.e. that there will be only child.size.sectors between
any two consecutive child.start_sector items).
That's correct. The shadow code is responsible for validating each
insert of a Slice or Partition object. TI/TD is meant to be quite
stupid. If the consumer doesn't check the errsvc, they may miss an
overlapped child. get_gaps() is there as a convenience method for
consumers.
You could sort self._children on child.start_sector and get rid of the
V_BACKUP child, then check the "child" pairs for overlap easily.
Additionally you can get rid of the usage list entirely, which would
simplify the loop at 255-267. In fact, the 241-246 and 255-267 loops
could probably be combined and simplified together.
The get_gaps() methods have been rewritten a few times now (with another
iteration to come). Take a look when we send out the second round of code.
243: Is there such a thing as a backup partition? (I know about a
backup slice.)
No there's not.
Thanks, Jack!
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
