Could OGC API - Processes - Part 1 v2.0 use the same PubSub approach as OGC API - EDR?
ghobona opened this issue · comments
This topic came up during the OGC October 2023 Open Standards Code Sprint.
The EDR API SWG is currently working on a PubSub extension for OGC API - EDR that is envisaged to form the basis for a OGC API-wide approach.
Given that OGC API - Processes - Part 1 includes a callback mechanism. There is a suggestion that the PubSub approach being taken by OGC API - EDR could be applied to the future version 2.0 of OGC API - Processes - Part 1: Core.
This GitHub Issue is intended to capture the views and discussion points on this topic.
The GitHub Issue can be closed once the SWG reaches a decision on whether OGC API - Processes - Part 1 v2.0 could use the same PubSub approach as OGC API - EDR. Ideally the decision would be formalised via a SWG Motion.
My view is that the Pub/Sub approach should be one option.
While it could potentially replace the simple callback option, the simplicity of the current callback probably justifies keeping it as its own requirements class.
The "polling" approach of the current asynchronous execution modes should also be preserved, since not all clients and use cases necessarily need or can easily support callback/notification mechanisms.
@ghobona @jerstlouis I think that the upcoming version of the specification should definitely offer AsyncAPI and the PubSub approach as a notification option in addition to the current polling and callback mechanism (which should be preserved for backward compatability).
@pvretano
Is there a sneak peek of what the PubSub might look like, and how it would integrate with the subscribers? Some sample JSON?
On my end, I have already extended the subscribers to allow emails on top of the callback URI. Would it be compatible with the proposed PubSub?
@fmigneault no samples yet. I'm still reviewing the EDR pattern before I start writing.
@fmigneault This is the discussion paper I believe:
13-NOV-2023: SWG concensus is to add this as a "additional" notification methodology ... not to replace what it already there and it would be its own conformance class.