Skip to main content

Minutes for WEBPUSH at IETF-95
minutes-95-webpush-1

Meeting Minutes Web-Based Push Notifications (webpush) WG
Date and time 2016-04-04 20:40
Title Minutes for WEBPUSH at IETF-95
State Active
Other versions plain text
Last updated 2016-04-07

minutes-95-webpush-1
WEBPUSH IETF 95

Wednesday Apr 4th 2016 1720-1920

Thanks to Mike Jones for taking notes.

*** *** *** *** *** *** ***
Action Items:

1. Core-draft #####
 - Brian : Update the draft so application server can ask (suggest) push server
 to
           to reuse existing subscription sets. Server will respond with 202 in
           this case.
 - Chair : After this is reflected, chair will ask if the draft is ready to go
 into WGLC.

2. Vapid-draft #####
 - Martin : Update the draft to allow dropping of messages to be conformant.
 - Chair : Will poll the list about adopting the draft.

*** *** *** *** *** *** ***

Original notes from Mike;

The Note Well was presented and blue sheets circulated

Brian Raymor described the current state of the WebPush core protocol
              The details are in his presentation
              He described changes in the Receipts logic

Martin Thomson expressed concerns about scalability of the receipt mechanism
and the number of listeners needed in clients
              Martin said that some of the streams may be privileged
              Martin described aggregated message scenarios
              Martin wants to allow applications to suggest that subscriptions
              be added to an existing subscription set
                             Brian: This would mirror the language we already
                             have for subscription sets Brian will create a
                             pull request for this for review on the mailing
                             list
Darshak Thakore: Asked if it’s a 202 or 200
              Martin: It’s a 202

Shida Schubert asked how many people had read the current draft
              About 8 had read it

Patrick Linskey: Ask about how we can use urgency and achieve prioritization
              Martin: We talked about this in the past

Martin Thomson presented about Vapid
              It enables applications to identify application servers using a
              JWT The details are in his presentation It lets request be
              correlated over time and for reputations to be established

Action Item: Martin to update to allow dropping messages to be conformant

              Vapid is currently mostly voluntary
              Some servers will eventually require it
                             Some application servers will require it to get
                             their messages through
              The real identity is the cryptographic one
                            Other attributes aren’t authenticated

Shida Schubert asked how many people have read the draft
              About 6 had read it
Brian asked how this affects milestones
              Given that it’s an optional feature, it doesn’t affect the
              current milestones

Martin: Reported on comments from Eric Rescorla (ekr) on interactions between
signing and encryption
              This may delay the content encoding work
Mike Jones: Asked how to resolve the signature/encryption issues
Martin: Said that the solutions are well-understood in general
              The solution will likely involve scoping operations until they
              can only be applied in ways that make sense It may involve a
              content encoding that does both

Patrick Linskey: Re-raised prioritization and achieving quality of service
              Wants a list of the things that we want to do with prioritization
              next Make sure that we’re not defining urgency in a way that
              doesn’t work for future use cases
Joe Hildebrand: Asked if we need more granularity
              Patrick: They internally currently use two dimensions: source and
              prioritization
Patrick:  Would like to be able to convey things like “after a day, a message
can be dropped”
              Someone mentioned that we already have a TTL
              Patrick is worried about the interactions between the features –
              not so much the features themselves
Joe asked if Patrick wanted to hold up last call for this
              Patrick responded that he didn’t want to hold things up but he
              wanted to work on this in parallel

Alissa Cooper: This suggests that people review the draft with respect to
extensibility
              Several parts of it are not currently extensible