2008/6/19  <[EMAIL PROTECTED]>:
>> This is my raw prototype (in other words, it has passed make test and
>> a few of my own personal tests, but it isn't yet polished):
>> http://cr.opensolaris.org/~swalker/pkg-2022/
>
> Meta-comments from me:
>
> * It would be great if we could find another way to pass information
>  about the intent instead of hard-coding in the functions as a string.

I could make them constants and put them in the misc.py package. Reactions?

> * Why is the protocol version included in the
>  per-authority image configuration.  Would you please explain why you
>  chose to do it this way?

It's not the protocol version, it's the protocol type :-)

Yes, I know that's splitting hairs.

Basically, I needed a place to store the protocol type and the
authority seemed the most logical since, at the moment, we don't have
mirrors and the setting is basically per-authority.

> * You need to cope with receiving timeouts in the touch_manifest() path
>  now.

Hrm.

> * Are we guaranteed to get a 501 error when sending this path?  I would
>  much rather we get something like a 404, and then fall back to the old
>  protocol.  This is going to cause confusion when we introduce content

The old depot server will return a 501 -- guaranteed. I obviously
can't change what it returns :-)

>  mirrors, since they're actually not going to support catalog and
>  manifest operations at all.  In that case, it seems appropriate to
>  return a 501, since the method isn't supported on the server side at
>  all.  Whereas in this case the method is implemented, but you can't
>  find what was requested.

The new depot server should return a 404 if you attempt a HEAD on a
manifest that doesn't exist.

Can you clarify a bit?

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

Reply via email to