On Wed, 2010-05-12 at 14:16 -0700, Danek Duvall wrote:
> > On Wed, 2010-04-28 at 16:42 -0700, Danek Duvall wrote:
> > > Tim Foster wrote:
> > > 
> > > > http://cr.opensolaris.org/~timf/pkgdepend-smf
> > 
> > > smf_manifest.py:
> > > 
> > >   - line 236ff: why not just
> > >     
> > >         try:
> > >             fmris = set(get_smf_dep...)
> > >         except:
> > >             ...
> > 
> > That's ok, so long as I set fmris before the try..except clause.
> > (more below)
> 
> I think we both meant "dep_fmris" there, but yes, I see your point.  Still,
> you don't need the if.

Yep, I had removed it there.

> But why show it twice?  Or are you simply dealing with the possibility that
> resolve_smf_dependency() returns None, in addition to the one where it
> raises a ValueError?

You're right - resolve_smf_dependency() can't return None, I've
consolidated this into a single elist addition.

> > I was following the style in pkg-gate/doc/tags-and-attributes.txt - we
> > should probably do a clean up of that documentation if underscores are to
> > be discouraged

> Alan's working on a bit of a cleanup of that document, but I hadn't asked
> him to do that; perhaps that's worthwhile.  There are a bunch of existing
> attributes with underscores which will be a bit tricky to change, and a
> bunch of proposed attributes with underscores which we can just avoid.

Great stuff. Just pointing it out.

> > > importer.py:
> > >   - line 174: once this if triggers, you can break out of the loop.
> > Won't it complain if I try to alter a variable I'm currently iterating over?
> 
> You're not modifying the object you're iterating over, you're just binding
> a different object to the same name and then immediately breaking out of
> the iteration.  I expect it'll be just fine.

Yep, you're right.

> > > It might be a useful enhancement to allow search to find appropriate
> > > subsets of SMF FMRI names.

> > Arguably, a user could do this already using pkg search wildcards
> >  - do you think it's worth complicating search with specifics about SMF?
> 
> Not sure, though I think that when people are used to typing "svcadm
> restart nwam" or "svcadm enable -t network/physical:default", they'll be a
> bit annoyed if those don't just show up in search.  But until we devise
> better guidelines, we can suggest wildcarding as a workaround.

Ok.

Now that Shawn's file-based repository client access bits have landed,
I've also relaxed the check for file repositories in importer.py. 

I've got an incremental webrev with these changes at:

http://cr.opensolaris.org/~timf/pkgdepend-smf-v3-vs-v2

        cheers,
                        tim


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

Reply via email to