Re: Ordering

[Manoj Srivastava <srivasta@datasync.com>]

Hi,
>>"Jason" == Jason Gunthorpe <jgg@gpu.srv.ualberta.ca> writes:

Jason> I am going to bed and will read this long email in more detail
Jason> tomorrow, but it occured to me that pkg-order doesn't have to
Jason> worry about configuration ordering.

	True enough. 

Jason> Consider - you arrange packages so that dpkg will always unpack
Jason> them correctly, dpkg itself performs the configuration ordering
Jason> on the result.

	Correct. But the order I pass in to dpkg has never been
 different from the one that dpkg used for the configuration  (umm, in
 alll cases I have seen -- but I have not exhaustively tested that)

Jason> I am doing -BOTH- and that is why there must be two
Jason> stages.
	
	Two stages, maybe -- but not necessarily two _different_
 topological ordering. 

Jason> Also remember I am looking at this from the view of the
Jason> output of a simulator, I know exactly when the rules are
Jason> broken, how many breaks are generated and so on.

	Yes, that is a truly neat concept. I have to steal that for
 pkg-order. 

Jason> Maybe we should just implement your routine exactly as is.. I
Jason> was never too comfortable working from that short synopsis of a
Jason> possible algorithm..

	;-)

	A new version of pkg-order is on its way, quite possibly the
 final patch required (there are a few bells and whistles I want to
 add, but ... it shall rapidly be obsolete, so I may not bother.

	manoj
-- 
 Oh what a tangled web we weave, When first we practice to deceive!
 Sir Walter Scott
Manoj Srivastava  <srivasta@acm.org> <http://www.datasync.com/%7Esrivasta/>
Key C7261095 fingerprint = CB D9 F4 12 68 07 E4 05  CC 2D 27 12 1D F5 E8 6E