Shawn Walker wrote:

> On 07/19/10 10:55 AM, Danek Duvall wrote:
> >Tim Foster wrote:
> ...
> >>I also ran pkglint over the manifests, giving
> >>
> >>ERROR pkglint.manifest003         variant.arch not declared in 
> >>pkg:/developer/opensolaris/[email protected],5.11-0.144
> >>ERROR pkglint.manifest003         variant.arch not declared in 
> >>pkg:/system/trusted/[email protected],5.11-0.144
> >>
> >>Should we be reporting the variant.arch errors for these?  I know they're
> >>group packages, so perhaps this lint check needs to be changed only
> >>complain if the package delivers actions other than 'set' and 'depend'.
> >
> >If the package is identical across all instances of a variant, why should
> >that variant be declared?  I'm willing to put those in, but it seems a
> >little odd.
> 
> I believe Bart has mentioned to me before that a package doesn't have
> to declare its variants.  Certainly, I don't believe the solver cares
> in this scenario.

Indeed; Bart confirmed that having the variant.arch set action is
unnecessary; the merge program will add them in for us.  So for
consistency, I'll pull them out entirely, unless someone can think
of a good reason to have them in.

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

Reply via email to