Brock Pytlik wrote:
p5i.py:
Lines 161-164: I don't really understand this comment.
In short, the output data structure is a giant dict that looks something
like:
publishers: {
'foo': {
'packages': ["bar", "baz"],
},
},
...so because of that, is not only unnecessary, but undesirable to embed
the publisher information into the FMRI.
194-203: is there any reason we might want to keep the items in this
list associated with their particular repositories? maybe they are and I
just don't realize it, but I'm confused as to how, for example, a mirror
would work w/out knowing which repo it mirrored
I'm a little lost as to what you mean by "associated with their
particular repositories". The code is adding an array of repositories
that belong to a specific publisher underneath the publisher's dict entry.
server/depot.py:
line 53: is there a way to do this w/out importing a client module? So
Not currently; pkg.client.publisher really should have been
pkg.publisher, that was my mistake.
far, we've had a very clean separation between client-only, server-only,
and shared code. I'd like to continue this going forward as it'll make
splitting SUNWipkg up into client and server packages much easier. If
this is a hack to keep the level of change low and get it in before the
gate closes, I'm fine w/ that but please file a bug then to refactor the
code as necessary.
Essentially. pkg.client.publisher should really be pkg.publisher and
pkg.client.api_errors needs a lot of its errors moved to pkg.api_errors
as they are often common between server and client.
896: couldn't this be 'if "*" not in pfmri'?
if "*" in pfmri, actually. Thanks.
900: I think the try block from 876-884 could be moved inside here
Not quite. My intent was to abuse MatchingFmri to validate that the
partial FMRI provided (such as 'SUNWvim') could actually be a valid FMRI
(that is, the pkg_name portion).
Filing a bug to remove the hacks related to packagemanager's requirement
of the p5i file extension would be nice.
They've already done that.
Cheers,
--
Shawn Walker
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
