Package dependency analysis

[Mark McLoughlin]

Hi David,

On Wed, 2006-12-06 at 12:16 -0800, David Lutterkort wrote:
> 
> I am not sure that graphical repesentations of dependencies are all that
> useful. I'd rather have a (GUI based) tool that is focused on answering
> specific questions; for example,
> 
>       * given a set of packages/groups, what is the package set that
>         anaconda will install with that input (i.e. closure under dep
>         solving)
>       * given a set of packages/groups I, and its closure C, why is
>         package X in C ? This might actually benefit from showing the
>         full dependency path from the initial packages to the resolved
>         set, though just highlighting the member(s) of I that cause X to
>         end up in C might be enough
>       * given the sets I and C and a specific package X in I, which
>         packages in C are pulled in by X ? Which ones are pulled in just
>         by X and which ones by X and other packages in I ?

	That's more or less what I'm getting at :-)

	I've done a little hacking to make it more obvious what it is I'm
trying to do. Give it a whirl ...

  $> wget http://people.redhat.com/markmc/depsgraph/depsgraph-0.31.tar.bz2
  $> rpmbuild -ta ./depsgraph-0.31.tar.bz2
  $> rpm -Uhv /usr/src/redhat/RPMS/i386/depsgraph-0.31-1.i386.rpm

	Assume you're looking at kernel deps in FC6:

  $> depsgraph http://download.fedora.redhat.com/pub/fedora/linux/core/6/i386/os/ kernel

	Scroll down through the list looking for anything unusual. Notice
python, double-click on it. [1]

	"Ah, something called cracklib requires it" ... click on cracklib. [2]

	"Ah, pam needs that ... hmm, cracklib probably shouldn't require python
then" [3]

Cheers,
Mark.

[1] - http://people.redhat.com/markmc/depsgraph/screenshots/depsgraph1.png
[2] - http://people.redhat.com/markmc/depsgraph/screenshots/depsgraph2.png
[3] - http://people.redhat.com/markmc/depsgraph/screenshots/depsgraph3.png