Shawn Walker wrote:

> On 04/13/10 05:45 PM, Danek Duvall wrote:
> >Shawn Walker wrote:
> >
> >>http://cr.opensolaris.org/~swalker/pkg-bp-11833/
> >
> >image.py:
> >
> >   - Are the changes to this method really necessary to fix 11833?  I'm
> 
> Which method?

Oops.  __filter_install_matches().

> >     guessing that this is an unfiled bug allowing ambiguously named
> >     packages to be installed if some other criteria is met.  But there
> >     aren't any real changes here, except in how it's all implemented, are
> >     there?  I changed this because I was adding the ability to disambiguate
> >     via obsolete state, but you don't have that here.  Was there some other
> >     bug here that I fixed, or that you'd fixed in your catalog v1 wad?
> 
> In the 2009.06 branch, the same filtering routine is used for install,
> uninstall, and list matching.

I only see __filter_install_matches() being called in make_install_plan.
Are you talking about __multimatch() as used via inventory()?

> I thought the fix I was doing was to fix the ambiguous install case
> for sunstudio packages by fixing the ambiguous list case since both
> use the same code.  Fixing both at the same time seemed logical.

Fixing both is fine, but I don't see how the changes to
__filter_install_matches() are involved.

> >t_pkg_list.py:
> >
> >   - Should we add these tests to the current branch, or is the same
> >     functionality tested in some other way already?
> 
> No, these should probably be added.  But in trunk and 2010.x branch,
> install filtering is its own thing separate from list filtering.

Okay.  Please make sure a bug gets filed to add these cases in the current
branch.

Thanks,
Danek
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to