Stream: ig publishing requirements
Topic: Invariants
Grahame Grieve (Sep 16 2021 at 05:19):
If you look at this page, you'll see a long list of useless invariants: https://build.fhir.org/ig/HL7/fhir-shc-vaccination-ig/StructureDefinition-shc-patient-general-dm.html
Grahame Grieve (Sep 16 2021 at 05:20):
that's because the editor hasn't set show-inherited-invariants = false in their IG parameters. And showing all the invariants defaults to true
It should default to false, not true. Does anyone object if I make this change?
Jose Costa Teixeira (Sep 16 2021 at 06:46):
That has a big impact on all IGs. Would it be possible to use the process as we defined for the template? Especially the part of a few weeks for feedback and putting it in current for at least 1 month
Jose Costa Teixeira (Sep 16 2021 at 06:47):
I think this change can also be made in the base template. Would that be an option?
Grahame Grieve (Sep 16 2021 at 11:17):
that would be an option. I just thought we already agreed it would be default. And it's not that much a change since everyone can still have the old way if they want
John Moehrke (Sep 16 2021 at 11:23):
I had no idea this could be done... so count me as someone who would benefit from this default change. I would benefit both as an author and a reader... as a reader of specifications I stop reading when I see the invariants because of the volume they are just gibberish to me.
Jose Costa Teixeira (Sep 16 2021 at 11:23):
i didn't recall if that was supposed to be the default.
Jose Costa Teixeira (Sep 16 2021 at 11:24):
(and I'm usiug this to exercise the process, so that the process actually helps)
Jose Costa Teixeira (Sep 16 2021 at 11:24):
I also don't like to see those invariants
Jose Costa Teixeira (Sep 16 2021 at 11:25):
btw, in Confluence the default is still to show inherited variants https://confluence.hl7.org/display/FHIR/Implementation+Guide+Parameters
Grahame Grieve (Sep 16 2021 at 11:39):
yes no one reads the doco ;-)
Lloyd McKenzie (Sep 16 2021 at 16:36):
There's already a proposed change to handle this differently:
- re-organize the table so that it has one row per invariant, not one row per element
- split into three different tables with one per tab, so that 'differential' only shows invariants that are different, 'mustSupport' only shows invariants on mustSupport elements, and 'snapshot' shows all invariants that are in play.
In snapshot, you really do want to see the inherited invariants too, but the first change will make the invariants that are in effect everywhere much less onerous to read.
Grahame Grieve (Sep 16 2021 at 18:32):
and where is that proposed change in the process?
Jose Costa Teixeira (Sep 16 2021 at 19:18):
https://github.com/HL7/fhir-ig-publisher/issues/272 ?
Jose Costa Teixeira (Sep 16 2021 at 19:21):
Since we don't have a kanban yet :stuck_out_tongue_wink: I guess this is in step 2 or 3
Lloyd McKenzie (Sep 16 2021 at 19:50):
It's been surfaced here and discussed both here and on the call. I think it's now waiting to be applied. I expect to get to it end of this month or early October.
Grahame Grieve (Sep 16 2021 at 21:34):
just for interest, where's the record that it was discussed? What's the procedural expectations here?
Lloyd McKenzie (Sep 16 2021 at 23:29):
I was wrong. Brought it to public discussion before we started formal calls. Full set is here: https://chat.fhir.org/#narrow/stream/179294-committers.2Fannounce/topic/Proposed.20IG.20template.20changes
Lloyd McKenzie (Sep 16 2021 at 23:30):
So well before we had procedures at all. I know @Eric Haas wants to cover at least one of these during our timeslot at the WGM. I think we're in the "finalize agreement on proposed design" phase?
Jens Villadsen (Sep 17 2021 at 21:49):
@Chris Moesel you might wanna add 'show-inherited-invariants' to the exhaustive example on https://fshschool.org/docs/sushi/configuration/#exhaustive-example
Chris Moesel (Sep 20 2021 at 13:19):
@Jens Villadsen - noted! https://github.com/FSHSchool/site/issues/50
Jens Villadsen (Oct 07 2021 at 13:03):
@Chris Moesel - I think your show-inherited-invariants: false (https://github.com/hl7dk/dk-core/blob/f22c4d74ef37fe276fcd8e32d43c511fe1f583bf/sushi-config.yaml#L29) is a bit buggy. In my master branch where I show inherited invariants, I see them all. When I disable it, all invariants are hidden - including those that shouldn't be. Take for instance this: https://build.fhir.org/ig/hl7dk/dk-core/branches/master/StructureDefinition-dk-core-cpr-identifier.html#constraints vs. https://build.fhir.org/ig/hl7dk/dk-core/branches/1.1.0/StructureDefinition-dk-core-cpr-identifier.html - the invariant named cprwhich is defined in this profile is also hidden which it shouldn't be (and it can still be found within the structuredef)
Lloyd McKenzie (Oct 07 2021 at 13:05):
The intention is to retire that flag and instead move invariants into the 3 views - snapshot, differential and mustSupport (and also to simplify the table so it's less verbose)
Chris Moesel (Oct 07 2021 at 13:07):
@Jens Villadsen - SUSHI just passes that parameter on to the IG Publisher. SUSHI doesn't actually do anything to hide invariants itself. So, if there is a bug there, it's actually in the IG Publisher.
Jens Villadsen (Oct 07 2021 at 14:40):
@Lloyd McKenzie now logged at https://github.com/HL7/fhir-ig-publisher/issues/321
Jose Costa Teixeira (Feb 03 2022 at 19:55):
is there anything we can do on this bug https://github.com/HL7/fhir-ig-publisher/issues/321 ?
Jose Costa Teixeira (Feb 03 2022 at 19:56):
I'm happy to try and make a PR if I find the code...
Jose Costa Teixeira (Feb 03 2022 at 19:56):
(I wanted to demonstrate invariants and the noise with inherited invariants makes it impractical)
Elliot Silver (Feb 03 2022 at 20:47):
I think this was covered by the discussion on yesterday's call regarding J#20385, but it may not be fully covered. That issue has to go to FHIR-I.
Grahame Grieve (Feb 03 2022 at 21:08):
really? how is that linked?
Elliot Silver (Feb 03 2022 at 21:11):
We discussed a new extension to suppress display of invariants, although the context was “inherited invariants made redundant through further profiling”
Grahame Grieve (Feb 03 2022 at 21:12):
ok so that's something slightly different
Lloyd McKenzie (Feb 03 2022 at 21:12):
There's a different tracker that should eliminate the need for the "show inherited invariants" flag entirely (which will make moot whether it works or not). But that's not 20385.
Jose Costa Teixeira (Feb 03 2022 at 21:57):
Actually there's no noise as I thought there was (I made a oopsie in my IG). But the bug is still there:
http://build.fhir.org/ig/hl7-be/hl7-be-patient-dossier/StructureDefinition-be-vaccination.html
The imm-1 is not declared in this example
Last updated: Apr 12 2022 at 19:14 UTC
