Hi All,

We have submitted an update to the ALTO Protocol draft which addresses
the outstanding items from the last IETF and the couple of discussions
on the list.

The main portions of the protocol have remained relatively stable for
quite a while now, so it would be good to look towards wrapping this
up.  Based on recent activity, I know folks are already looking
towards how we want to extend it.

The changes in this version are:
- Removed the section on redistribution (since some of the mechanism
there may eventually be provided by the JOSE WG).  The intent is to
submit that as a separate draft, but that isn't done yet.

- Removed the Server Information Service, since its only existing
purpose was to carry certificates used in the now-removed
redistribution mechanism.  It didn't seem worthwhile to have an
interface in the base protocol that was entirely unused, and it was
independent from everything else and thus would fit well in a separate
document. However, I'd be happy to add that back if the WG feels like
we should have that in the base protocol.  Discuss :)

- Added an explicit note that costs in the cost map are end-to-end
costs.  Let us know if the added text still does not make the
semantics clear.

- Changed the data type of costs from JSONNumber to JSONValue in the
formal specification. Implementations are required to use this as
JSONNumber unless an extension document indicates that something else
is permitted (and that extension document should indicate when and how
to do this without breaking existing implementations).  This was
provided as an extension hook for cost values that are not numbers.
If there are ideas for how this text could be refined, that would be
great.

- Added greater than or equal to, less than or equal to, and equal to
operators to constraints.

- Added an explicit note to caution clients against constraints
applied to ordinal cost maps.  Based on Bill's thread "What do
constraints mean for ordinal costs?", the cleanest solution seemed to
be to indicate that cost constraints shouldn't pertain to ordinal cost
maps. Looking at this again, it seemed unnecessarily-constraining to
forbid it (meaning, it would be an error to do so), but there is now a
note indicating why clients may not get what they asked for.  If that
isn't sufficient, please let us know.

Thanks!
Rich
_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto

Reply via email to