Bart Smaalders wrote:
> http://cr.opensolaris.org/~barts/13241/
You're fixing bug 13968 here, too.
Does it make sense for a "normal" package (one without any incorporate
dependencies) to set this attribute?
pkg_solver.py:
- line 712,965,982: remove trailing spaces
- line 981: is there any reason to think we might have more than one
value for this attribute?
- line 994: This allows for "core-os" to release a hold by
"core-ostracized". Do you want to restrict it to the dotted hierarchy?
[r for r in relaxed_holds if hold == r or hold.startswith(r + ".")]
- line 996: does this imply that we want an audit check at some point
that warns when you've set install-hold on a package that also has a
versioned dependency on another package? I'm also not sure how the
code following this has anything to do with non-incorporation packages.
I think this comment could use some more explanation. The set
arithmetic needs some English backing it up -- I think probably because
there are quite a number of negatives here, and piecing them all
together is difficult in my head.
So, versioned_dependents is the set of packages which have versioned
dependencies on some package. From this set, you're now removing the
packages which have the install-hold property and haven't been relaxed.
That means it's going to have basically all normal packages as well as
all incorporations which have had their install-hold property relaxed.
That means what we return will have those relaxed incorporations
removed from the list of incorporations.
I'm also wondering if versioned_dependents also shouldn't just
accumulate incorporation packages instead of all packages, since its
final use is only in comparison with the incorporations. What am I
missing?
- line 1004: "incorportation" -> "incorporation"
Is it worthwhile testing that installing a package constrained by an
incorporation (and having a versioned dependency on that incorporation) as
well as the incorporation itself on the commandline works?
Thanks,
Danek
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
