Re: package dependency as applied math (was: Re: tape deity)
Hi,
>>"John" == John Hasler <john@dhh.gt.org> writes:
John> Jason Gunthorpe writes:
>> ...you may -NEVER- have predepends looping or predepends+depends
>> looping which means it is always possible to create an order where
>> all dependandants (and their dependants) of all predepends come
>> before the predepend.
John> Has anyone attempted a mathematical model of the general
John> dependency problem? It looks like a graph theory problem;
John> perhaps there is already a general solution.
Of course. It is a simple topological sort. How do you think
that the ordering code works? pkg-order, and now deity, anre
not haphazard trial and error methods.
The issue is more complex than simple graph theory algorithms;
since diffrent actions requirte different set of edges in the
graph; also, not all nodes are equivalent, and not all edges
have the same cost.
Consider the invariants that must never be broken to retail a
working system, and you have a gliimering of the complexity that is
involved.
manoj
we are not all code first, think later genre of hackers, you know
--
The inability to benefit from feedback appears to be the primary
cause of pseudoscience. Pseudoscientists retain their beliefs and
ignore or distort contradictory evidence rather than modify or reject
a flawed theory. Because of their strong biases, they seem to lack
the self-correcting mechanisms scientists must employ in their
work. Thomas L. Creed, "The Skeptical Inquirer," Summer 1987
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
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: