20190304 Community Meeting Minutes

Time: 19:00-20:00 UTC (20:00-21:00 CET, 14:00-15:00 EST, 06:00-07:00 AEDT) Location: Join Zoom Meeting https://zoom.us/j/573322038

See https://youtu.be/dcUSUVSP7z8 for the recording of this meeting

R4/1.2-beta

We will remain in beta phase for a bit, ironing out performance and bugs.
Working on stabilizing POCO-less parsing
Will need to make sure validation of decimals pre-R4 picks up on the (back then) unsupported exponention notation, see https://github.com/FirelyTeam/fhir-net-api/issues/898.

Rewrite of the FhirPath engine. FHIRPath->Linq.Expression. Compile().
Performance improvements. Patient.a.b.c.d.e.f(g.h.i) is particularly bad, re-running the focus as IEnumerable on every step.
Making configuring validation errors more fine-grained (as discussed before)
George is working on isolating poco-dependent code/"common" code base, so we have more shared code across branches.
- We can use the Resource class across all the versions. Yes!!!
- George created Roslyn codeanalyzers to signal features that are incompatible with the fhir version chosen.
- E.g. if you are in DSTU2 and you use Resource.meta.source, you get a warning.
- This is scheduled for 2.x
- Generating IStructureDefinitionSummaryProvider is a nice solution too. Would fix the problem of the order/insummary changes per fhir version

Luke is on the call to plead his case for .NET4 support
It's simple to add a new target, but compilation times suffer. We should have a DEBUG/RELEASE config for TargetFrameworks
Luke will try to see if the PR for the new FhirClient is doable in 4.x
Rediscuss on april 1st.

% and EvaluationContext - add a way to add your own variables
Working being done is around rewriting the expression tree to flag lambdas (like where), and adding support for lambdas:
- FP N1: substing(4) where(\this => family = 'xxxxxx' and $this > 'b')
- aggregate(($this,$index,$total) => )
- define new functions! 'let $elevenProof = ($a) => ................. '
Also, we can pull more stuff done currently at runtime into static analysis phase, like resolving $focus.
Decided to keep the current tree returned by FhirPathCompiler.Parse() unchanged. Tree rewriting is done before compilation.

We decided to stick with just PermissiveParsing true/false for 1.2, a proper solution takes more time. Don't want to introduce temporary fixes which we'd need to keep on supporting.
Not before summer

We moved the FHIR .NET API to https://github.com/FirelyTeam/fhir-net-api
Brian will do an experimental run on code generation of an "unknown code" in all enums.
Brian has started to pull out the Web.Api stuff from the repo to new repo https://github.com/brianpos/fhir-net-web-api
Luke will try to see if the PR for the new FhirClient is doable in 4.x
Ewout to set up a targeted meeting to discuss these different interfaces for WebApi
Ewout to schedule a meeting to show proposed changes to the internals of the validator [DELAYED until further notice]