FHIR Chat · Discussion on Subscriptions / Messaging

Stream: fhir/infrastructure-wg

Topic: Discussion on Subscriptions / Messaging

Josh Mandel (Feb 24 2020 at 22:26):

We're looking for a block of time next week to continue talking about subscriptions / messaging.

Please fill out this doodle poll

FYI @Gino Canessa @Grahame Grieve @Eric Haas @Lloyd McKenzie @Vassil Peytchev @Michael Donnelly @Rick Geimer @Christiaan Knaap @Bas van den Heuvel from discussion today.

CC @FHIR-I Co-chairs

Josh Mandel (Feb 24 2020 at 22:29):

Hold on, updating the poll to reflect NEXT WEEK.

Josh Mandel (Feb 24 2020 at 22:31):

~~Fixed!~~

Josh Mandel (Feb 24 2020 at 22:38):

Now fixed for real.

Vassil Peytchev (Feb 24 2020 at 22:46):

@Gino Canessa probably needs to provide a post-fix response...

Gino Canessa (Feb 24 2020 at 22:48):

Yep, thanks!

Josh Mandel (Feb 27 2020 at 20:01):

Okay, the "Subscriptions & Messaging Discussion" is scheduled for March 4 at 2p ET. The Web link for the Teams call is https://aka.ms/fhir-subscriptions-call
and the entry in HL7's conference call tracker is here.

Thanks for indicating you can make it, @Bas van den Heuvel @Christiaan Knaap @Michael Donnelly @Grahame Grieve @Lloyd McKenzie @Vassil Peytchev @Eric Haas @Gino Canessa .

Josh Mandel (Feb 27 2020 at 20:02):

HL7 also generates a handy ics calendar file here.

Josh Mandel (Feb 27 2020 at 20:09):

(Do NOT dial into the HL7 phone number at "+1 770..."; follow the web meeting link or the "join Teams by phone" instructions.)

Josh Mandel (Mar 04 2020 at 18:22):

Reminder: our call is in 38 minutes! Relevant links:

https://aka.ms/fhir-subscriptions-call to join (send a message here if you have trouble)
Slide deck to guide discussion

Michael Donnelly (Mar 04 2020 at 19:15):

If you're using the deck again in the future, Josh, if you change "Ability to convey more event-related content" to "Ability to convey additional metadata," I think you'll be less likely to get concerned questions.

Josh Mandel (Mar 04 2020 at 20:01):

Hmm, I think "metadata" confuses me here -- but I've updated the langauge slightly in this direction.

Josh Mandel (Mar 04 2020 at 20:04):

Thanks for today's call! I'll upload a video link when the recording is ready. Deck is here. Meanwhile, we've got three recommendations that I'll post in discussion threads here for thumbs-up, thumbs-down, and feedback.

Josh Mandel (Mar 05 2020 at 03:40):

Video link from today's call: https://youtu.be/25WnxO-BPmg (may take a few minutes to appear)

Eric Haas (Mar 05 2020 at 16:53):

So we discussed a more radical alternative at the end of the call: based upon @Vassil Peytchev discussion and @Gino Canessa mockup. (See the video). Throw off the yoke of Bundle and create a Notification resource through which all notifications would funnel whether they are explicitly sub-pub or "unsolicited" and historically used V2 messaging or not. Then it would be clear that FHIR messaging is only relevant for Request and response ( e.g. orders) messaging. This approach has the attraction of:

solving the issue of defining boundaries between messaging and R5 subscription+REST
a simpler (IMO) unified approach
carves out a distinct suite of resources for subscriptions for all channels.

I think it deserves more discussion.

Vassil Peytchev (Mar 05 2020 at 18:15):

Yes, there needs to be more discussion on all of these topics. It is worth noting that whether or not a Notification "throws off the yoke of Bundle" or not is orthogonal to its use for both subscription notifications and unsolicited notifications/implicit subscriptions. The latter could/should inform the former, but it is the former that is under discussion. The current proposal is to stay with Bundle.

Eric Haas (Mar 05 2020 at 19:25):

I agree it is orthogonal, but ,at least for me, the light bulb went off in my head when this came up.

Josh Mandel (Mar 05 2020 at 20:23):

It's not clear to me that FHIR-I can/would stomach the idea of creating new (non-Bundle) ways to convey a list of resources inline. I'd rather keep that discussion orthogonal indeed, so we can make progress toward deciding on the recommendations above. Given the robust discussion on yesterday's call, I'm surprised/concerned that we're not seeing any discussion (or endorsement or rejection) of these recommendations here.

Eric Haas (Mar 05 2020 at 21:05):

This issue I have is does endorsing these preclude consideration of the more radical approach.

Paul Church (Mar 05 2020 at 21:20):

Thanks for the recording, I wasn't able to make that meeting time.

Gino Canessa (Mar 05 2020 at 21:43):

Eric Haas said: