Parted 1.6.0-pre3

[Andrew Clausen]

Hi all,

I'm back home in Australia :)  OTOH, uni is starting, but usually
the first 1-2 months aren't very time-consuming, so I should be
able to sort out the mess.

I made a fairly big change (about a month ago, just I've been
unlucky with my computer systems since then), so that rather than
verifying that partitions are not overlapping, an overlap constraint
is generated and applied (along with all the other constraints).

This is really the only sane solution to this problem:
(1) user provides a geometry that is close to overlapping
(2) *_partition_align() algins the geometry such that it overlaps

In practice, (2) didn't happen much.  Now it does, due to:
(1) auto-gobble-up of small amounts of free space
(2) some special cases with weird/unknown bios geometries, etc.

This Parted is broken, in the sense I haven't fixed the regression
tests.  i.e. If you attempt to make an overlapping partition,
Parted will "fix" it (and check if that's ok), rather than say
it can't be done.

An unfortunate side-effect of all of this is when it's impossible
to satisfy all constraints, you just get a "unable to find
a solution" error message, not an error message like "too small"
or something.  It seems like a difficult problem to solve well.
Ideas?

(I think somethings can be done, like if the constraint has
min_size > max_size...)

Andrew