On Wed, Aug 15, 2012 at 06:50:42PM -0700, Brock Pytlik wrote:
> On 08/07/12 19:48, Edward Pilatowicz wrote:
> >webrev:
> >https://cr.opensolaris.org/action/browse/pkg/edp/pkg.refresh/webrev/
> >
> >bugs:
> >7187946 pkg running in a zone always thinks it's modifying a non-active BE
> >7185502 need to document pkg change-{variant|facet} --no-refresh options
> >7173792 pkg doesn't auto-refresh previously used and disabled sysrepo 
> >publishers
> >
> >thanks,
> >ed
> >_______________________________________________
> >pkg-discuss mailing list
> >pkg-discuss@opensolaris.org
> >http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
> image.py:
> 977: Why's it necessary to set the publisher's meta root here?
> Alternatively, why do only these publishers need their meta root
> set?

the reason we set last_refreshed to None is because we want the
last_refreshed set method remove the on-disk last_refreshed cache file.
but if meta_root is not set then that method doesn't know where the file
is so it won't actually delete it (it will only update the in-memory
state).

> 978: Personally, I think this line really belongs in
> imageconfig.__merge_publishers. Couldn't we add it in as part of the
> step of generating modified_pubs around line 1286 in imageconfig.py?
>

i tried to do this in a previous iteration and it didn't work out too
well.

the problem was that to do this we need to pass meta_root to the
BlendedConfig init routine (for the reasons mentioned above).  but
BlendedConfig objects can get allocated at two places, one of which
happens before we've actually determined the image version number.  but
the image version number is required to determine the meta_root path.  i
tried eliminating the other (earlier) place where we could allocate a
BlendedConfig object, but i couldn't get that working...

> t_pkg_sysrepo:
> 456: Why aren't we destroying the image here any more? After this
> loop is finished, we'll now have an extra random image laying around
> on disk (which may confuse developers using the -a option, though
> I'd need to see what kind of archive actually is produced).
>

fixed.

> You might consider adding tests where the user changes the value of
> the image-property "use-system-repo" from true to false, then back
> to true, and while it was set to false, origins were added or
> removed or packages were published to configured publishers.
>

i'll try doing this later today and see what i turn up.

ed
_______________________________________________
pkg-discuss mailing list
pkg-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to