[d-i] partkit vs autopartkit

[Joey Hess <joeyh@debian.org>]

There seems to be quite a lot of overlap between partkit and
autopartkit. I guess the former gives more control, and the latter
figures it all out for you (and can resize fat partitions). So we
probably want both, but for the immediate goal of getting d-i to work,
which should we focus on?

Overlap between the two includes device detection (not in partkit;
autopartkit uses ped_device_get_next() whatever that is, and selecting
which devices to partition. I think that partkit is more advanced there
as well, the table it displays looks pretty neat. I wonder if some of
this couldn't be broken off into a third udeb, call it partselect, that
handles picking the partitions. Or maybe it would make better sense to
merge the two code trees into one that builds 2 udebs, or to split it
off into a static library, or completly merge them, and offer the user
the ability to autopartition or manually partition each device in turn.
Thoughts?

-- 
see shy jo