Currently, we need to have specific constructors for each primitive, but we can also use the no-param constructors and then set the ObjectValue. This will allow us to keep track of invalid values too.

In the generated TypeScript_R4 the interface generation of ElementDefinitionExample has all the values (valueCode, valueBoolean etc) as mandatory so that the typescript checking fails when you try and instantiate an ElementDefinition with an example e.g
"example": [ { "valueUri": "http://hl7.org/fhir/v3/MaritalStatus", "label": "General" } ],
then the TypeScript type checking fails. This will be because the profile says that value[x] must have a value but there is no way to represent in an interface that I am aware of to say it must have only one of the multiple types of values that are allowed. If you mark them all as optional then it will type check correctly but not sure how your codegen could evaluate this pattern

This way you can have your own classes instantiated instead of the standard POCOs.

Is there still a requirement (for the UI or otherwise) to have an .index.json file? When loading, for example, https://simplifier.net/packages/nictiz.fhir.nl.r4.zib2020/0.9.0-beta.1 I get 0 resources, but do get some CodeSystems:

As reported here https://chat.fhir.org/#narrow/stream/179171-dotnet/topic/Microsoft.20fhir-codegen-cli/near/383005716

On first run, the download/caching code is bailing out with TryDownloadAndExtract <<< exception downloading: http://packages2.fhir.org/packages/hl7.fhir.r5.core/5.0.0-snapshot2/, One or more errors occurred. (Response status code does not indicate success: 500 (Internal Server Error).). Not sure if it's an upstream issue with the server, outdated URL etc.

`
➜ fhir-codegen git:(main) dotnet run -p src/fhir-codegen-cli
Warning NETSDK1174: The abbreviation of -p for --project is deprecated. Please use --project.
Starting FhirCacheService...
<<< using cache directory: /Users/preston/.fhir
<< cache contains 10 packges
Starting FhirManager...
LoadCached <<< Loaded and Parsed FHIR DSTU2
Loaded hl7.fhir.r2.core#latest as hl7.fhir.r2.core version 1.0.2
LoadCached <<< Loaded and Parsed FHIR STU3 with 0 errors and 5 warnings
Errors (only able to pass with manual code changes)
Warnings (able to pass, but should be reviewed)

• CodeSystem SNOMED_CT (snomedct): Duplicate proprety found: Laterality
• ValueSet Yes/No/Don't Know (yesnodontknow): No Version present
• ValueSet Example-inactive (inactive): No Version present
• ValueSet Common UCUM units (ucum-common): No Version present
• ValueSet All CPT codes (cpt-all): No Version present
  Loaded hl7.fhir.r3.core#latest as hl7.fhir.r3.core version 3.0.2
  LoadCached <<< Loaded and Parsed FHIR R4 with 0 errors and 1 warnings
  Errors (only able to pass with manual code changes)
  Warnings (able to pass, but should be reviewed)
• CodeSystem SNOMED_CT (snomedct): Duplicate proprety found: Laterality
  Loaded hl7.fhir.r4.core#latest as hl7.fhir.r4.core version 4.0.1
  GetPackageVersionsAndUrls <<< Failed to get package info: NotFound http://packages.fhir.org/hl7.fhir.r4b.core
  GetPackageVersionsAndUrls <<< Failed to get package info: NotFound http://packages.fhir.org/hl7.fhir.r4b.expansions
  LoadCached <<< Loaded and Parsed FHIR R4B with 4 errors and 6 warnings
  Errors (only able to pass with manual code changes)
• CodeSystem ResourceSecurityCategory (resource-security-category): Status field missing
• CodeSystem ConceptSubsumptionOutcome (concept-subsumption-outcome): Status field missing
• CodeSystem StandardsStatus (standards-status): Status field missing
• CodeSystem ResourceValidationMode (resource-validation-mode): Status field missing
  Warnings (able to pass, but should be reviewed)
• CodeSystem SNOMED_CT (snomedct): Duplicate proprety found: Laterality
• ValueSet Yes/No/Don't Know (yesnodontknow): No Version present
• ValueSet ACMECholCodesPlasma (example-filter): No Version present
• ValueSet Example-inactive (inactive): No Version present
• ValueSet Common UCUM units (ucum-common): No Version present
• ValueSet All CPT codes (cpt-all): No Version present
  Loaded hl7.fhir.r4b.core#latest as hl7.fhir.r4b.core version 4.3.0-snapshot1
  GetPackageVersionsAndUrls <<< Failed to get package info: NotFound http://packages.fhir.org/hl7.fhir.r5.core
  TryDownloadAndExtract <<< exception downloading: http://packages.fhir.org/hl7.fhir.r5.core/5.0.0-snapshot2/, One or more errors occurred. (Response status code does not indicate success: 404 (Not Found).)
  <<< inner: Response status code does not indicate success: 404 (Not Found).
  TryDownloadAndExtract <<< exception downloading: http://packages2.fhir.org/packages/hl7.fhir.r5.core/5.0.0-snapshot2/, One or more errors occurred. (Response status code does not indicate success: 500 (Internal Server Error).)
  <<< inner: Response status code does not indicate success: 500 (Internal Server Error).
  FindOrDownload <<< unable to resolve a directive: hl7.fhir.r5.core#latest
  Unhandled exception: System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation.
  ---> System.Exception: Failed to retrieve hl7.fhir.r5.core#latest
  at Microsoft.Health.Fhir.SpecManager.Manager.FhirManager.LoadFhirCore(FhirSequenceEnum sequence, String version, Boolean offlineMode, Boolean officialExpansionsOnly, String ciBranch) in /Users/preston/Developer/git/fhir-codegen/src/Microsoft.Health.Fhir.SpecManager/Manager/FhirManager.cs:line 351
  at Microsoft.Health.Fhir.SpecManager.Manager.FhirManager.LoadPackages(IEnumerable1 packageDirectives, Boolean offlineMode, Boolean officialExpansionsOnly, Boolean loadDependencies, Boolean skipFhirCore, String ciBranch, List1& failedPackages, Boolean areDependencies) in /Users/preston/Developer/git/fhir-codegen/src/Microsoft.Health.Fhir.SpecManager/Manager/FhirManager.cs:line 495
  at FhirCodegenCli.Program.Process(String fhirSpecDirectory, String outputPath, Boolean verbose, Boolean offlineMode, String language, String exportKeys, String loadR2, String loadR3, String loadR4, String loadR4B, String loadR5, String languageOptions, Boolean officialExpansionsOnly, String fhirServerUrl, Boolean includeExperimental, String exportTypes, String extensionSupport, Boolean languageHelp, String packageDirectory, String packages, String ciBranch, String languageInputDir) in /Users/preston/Developer/git/fhir-codegen/src/fhir-codegen-cli/Program.cs:line 464
  --- End of inner exception stack trace ---
  at System.RuntimeMethodHandle.InvokeMethod(Object target, Span`1& arguments, Signature sig, Boolean constructor, Boolean wrapExceptions)
  at System.Reflection.RuntimeMethodInfo.Invoke(Object obj, BindingFlags invokeAttr, Binder binder, Object[] parameters, CultureInfo culture)
  at System.Delegate.DynamicInvokeImpl(Object[] args)
  at System.Delegate.DynamicInvoke(Object[] args)
  at System.CommandLine.NamingConventionBinder.ModelBindingCommandHandler.InvokeAsync(InvocationContext context)
  at System.CommandLine.Invocation.InvocationPipeline.<>c__DisplayClass4_0.<b__0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass22_0.<b__0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass15_0.<b__0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass26_0.<b__0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass24_0.<b__0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<b__23_0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass21_0.<b__0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<b__8_0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<b__7_0>d.MoveNext()
  --- End of stack trace from previous location ---
  at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass11_0.<b__0>d.MoveNext()

`

Currently the deserializer is mostly concerned about the happy path. Add code that does the same checks as the current SDK does:

Xml syntactic checks

• incorrect/empty namespace . P, R:namespace ignored
• empty attributes . P, R: whitespace,null then value=null
• no attribute, no elements P ->S, R:empty child list, value=null SKIP?
• repeating elements inconsecutively P,->S
• Invalid FHIR Xhtml P ->T
• container resources with attributes . P, R:no contained found
• contained resources with attributes . P, R:no contained found
• container resources with multiple children . P, R:no contained found

Json syntactic checks

• incorrect use of null P, R: skip
• incorrect use/combinations of name and _name . P, R: return main
• incorrect use of arrays for name and _name . P, R:return main
• invalid json token type . P: skip unless some useful content found
• empty string values ->T, p:return null
• non-arrays with null values . P, R:return null
• empty objects . P, R:return null
• obsolete fhir_comments use . P, R:skip

There are just a small number of "patterns" when generating serializing/deserializing code, and some of these can be extracted into utility functions so we need to generate less code.

In the end, me might be able to swap the utility functions for calls to an interface so the same serializer/deserializer code can be used for xml and json.

The section https://github.com/microsoft/fhir-codegen#fhir-codegen-cli tells me:

To run this project from a command line:
dotnet run --project src/fhir-codegen-cli/fhir-codegen-cli.csproj -- [options]

However, running dotnet run --project src/FhirCodeGenBlazor/FhirCodeGenBlazor.csproj -- -p hl7.fhir.r4.core#4.0.1 --language TypeScript --output-path ~/R4.ts does not seem to generate any file.

As reported on https://chat.fhir.org/#narrow/stream/179171-dotnet/topic/Microsoft.20fhir-codegen-cli/near/383004416

See https://chat.fhir.org/#narrow/stream/179171-dotnet/topic/Microsoft.20fhir-codegen-cli/near/383004416

When running dotnet run --project src/FhirCodeGenBlazor/FhirCodeGenBlazor.csproj in the cloned project folder it won't automatically open a browser window. It is however running on http://localhost:5197 when I type this in the browsers.

This is on Mac with Arc as the default browser.

See: http://hl7.org/fhir/DSTU2/conformance.html

I think it should be ('xml','json')[] .. or string[] ?

I've succesfully generated the swagger files from my server but they are all in version 2.0. I could convert them one by one using the Swagger Editor but I was wondering if I can do it directly from th codegen.

Thanks!

It would be more ergonomic for string fields values such as these:

public static class AddressTypeCodes {
    public const string POSTAL = "postal";
    public const string PHYSICAL = "physical";
    public const string BOTH = "both";
  }

to be enums

Originally posted by @johnstairs in #2 (comment)

Hi,

I am trying to find a way to get a full OpenAPI / Swagger definition file for an Azure FHIR API Service.
I asked the azure support and they have redirected to this project.

I have tried to generate a swagger file from R4 definition but I am having difficulties.
If I try with the R4 from package with this command :
fhir-codegen-cli.exe --load-r4 latest --fhir-spec-directory c:\temp\fhirR4 --language OpenApi --output-path c:\temp\FhirR4.json

I can see that there is an error "duplicate property found : Laterality" and the output file seems missing objects like Account

Starting FhirCacheService...
 <<< using cache directory: C:\Users\sthomas\.fhir
 << cache contains 2 packges
Starting FhirManager...
GetPackageVersionsAndUrls <<< Package http://packages2.fhir.org/packages/hl7.fhir.r4.expansions contains NO versions
LoadCached <<< Loaded and Parsed FHIR R4 with 0 errors and 1 warnings

Errors (only able to pass with manual code changes)
Warnings (able to pass, but should be reviewed)
 - CodeSystem SNOMED_CT (snomedct): Duplicate proprety found: Laterality
Loaded hl7.fhir.r4.core#latest as hl7.fhir.r4.core version 4.0.1
+ c:\code\temp\R4.json
Done! Loading: 4.413s, Total: 5.025s

But If I try from the api fhir Service Url there is another error and there is no output :

fhir-codegen-cli.exe --fhir-server-url https://fhirwspocpublic-apiservice.fhir.azurehealthcareapis.com/metadata --language OpenAPI --output-path c:\code\temp\FHIRR4_cli.json
Starting FhirCacheService...
 <<< using cache directory: C:\Users\sthomas\.fhir
 << cache contains 2 packges
Starting FhirManager...
Requesting metadata from https://fhirwspocpublic-apiservice.fhir.azurehealthcareapis.com/metadata...
Connected to https://fhirwspocpublic-apiservice.fhir.azurehealthcareapis.com/metadata, FHIR version: 4.0.1
Server Information from https://fhirwspocpublic-apiservice.fhir.azurehealthcareapis.com/metadata:
            FHIR Version: 4.0.1
           Software Name: Azure Healthcare APIs
        Software Version: 2.3.60
            Release Date:
             Description:
               Resources: 146
GetPackageVersionsAndUrls <<< Package http://packages2.fhir.org/packages/hl7.fhir.r4.expansions contains NO versions
LoadCached <<< Loaded and Parsed FHIR R4 with 0 errors and 1 warnings

Errors (only able to pass with manual code changes)
Warnings (able to pass, but should be reviewed)
 - CodeSystem SNOMED_CT (snomedct): Duplicate proprety found: Laterality
Loaded hl7.fhir.r4.core#latest as hl7.fhir.r4.core version 4.0.1
Unhandled exception: System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation.
 ---> System.NullReferenceException: Object reference not set to an instance of an object.
   at Microsoft.Health.Fhir.SpecManager.Models.FhirServerInfo..ctor(FhirServerInfo source, FhirVersionInfo info) in C:\code\fhir-codegen\src\Microsoft.Health.Fhir.SpecManager\Models\FhirServerInfo.cs:line 139
   at Microsoft.Health.Fhir.SpecManager.Manager.Exporter.Export(FhirVersionInfo sourceFhirInfo, FhirServerInfo sourceServerInfo, ILanguage exportLanguage, ExporterOptions options, String outputPath, Boolean isPartOfBatch) in C:\code\fhir-codegen\src\Microsoft.Health.Fhir.SpecManager\Manager\Exporter.cs:line 136
   at FhirCodegenCli.Program.Process(String fhirSpecDirectory, String outputPath, Boolean verbose, Boolean offlineMode, String language, String exportKeys, String loadR2, String loadR3, String loadR4, String loadR4B, String loadR5, String languageOptions, Boolean officialExpansionsOnly, String fhirServerUrl, Boolean includeExperimental, String exportTypes, String extensionSupport, Boolean languageHelp, String packageDirectory, String packages, String ciBranch, String languageInputDir) in C:\code\fhir-codegen\src\fhir-codegen-cli\Program.cs:line 544
   --- End of inner exception stack trace ---
   at System.RuntimeMethodHandle.InvokeMethod(Object target, Span`1& arguments, Signature sig, Boolean constructor, Boolean wrapExceptions)
   at System.Reflection.RuntimeMethodInfo.Invoke(Object obj, BindingFlags invokeAttr, Binder binder, Object[] parameters, CultureInfo culture)
   at System.Delegate.DynamicInvokeImpl(Object[] args)
   at System.Delegate.DynamicInvoke(Object[] args)
   at System.CommandLine.NamingConventionBinder.ModelBindingCommandHandler.InvokeAsync(InvocationContext context)
   at System.CommandLine.Invocation.InvocationPipeline.<>c__DisplayClass4_0.<<BuildInvocationChain>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass22_0.<<UseParseErrorReporting>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass15_0.<<UseHelp>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass26_0.<<UseVersionOption>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass24_0.<<UseTypoCorrections>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<<UseSuggestDirective>b__23_0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass21_0.<<UseParseDirective>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<<UseDebugDirective>b__8_0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c.<<RegisterWithDotnetSuggest>b__7_0>d.MoveNext()
--- End of stack trace from previous location ---
   at System.CommandLine.Builder.CommandLineBuilderExtensions.<>c__DisplayClass11_0.<<UseExceptionHandler>b__0>d.MoveNext()

Can you please help me on what I am doing wrong ?

An alternate approach is to use something like this (https://github.com/microsoft/healthcare-shared-components/blob/master/src/Microsoft.Health.SqlServer/IndentedStringBuilder.cs and https://github.com/microsoft/healthcare-shared-components/blob/master/src/Microsoft.Health.SqlServer/IndentedStringBuilder.Generated.cs), where we keep track of the indentation level in the string builder.

Originally posted by @johnstairs in #2

I suggest doing _writer.Write(" "); _writer.WriteLine(value"); or similar to avoid unnecessary allocations

Originally posted by @johnstairs in #2

C# namespace should be more than fhir (and should be PascalCased).

Originally posted by @johnstairs in #2 (comment)

I think dates and times need to be represented as DateTime/DateTimeOffset

Originally posted by @johnstairs in #2 (comment)

It would be super useful if fhir-codegen could generate classes from any FHIR package. Right now it is limited to the core packages, but Gino mentioned this is not too far off?

I have heard multiple customers asking for this, to get them started with implementing a specification with strong typing/validation.

Previously discussed at https://chat.fhir.org/#narrow/stream/179171-dotnet/topic/Microsoft.20fhir-codegen-cli/near/383002869

QuestionnaireResponseGroup

This...

export interface QuestionnaireResponseGroup extends BackboneElement {
  /**
   * Reports can consist of complex nested groups.
   */
  group?: QuestionnaireGroup[];

should be this...

export interface QuestionnaireResponseGroup extends BackboneElement {
  /**
   * Reports can consist of complex nested groups.
   */
  group?: QuestionnaireResponseGroup[];

BTW - This was the only bug I found in R2 :)

Maybe we can extend the POCOs to have a backing dictionary to store unknown elements. Use a lazy dictionary?

I could imagine two variants of the codegen: one optimized for programmer-friendliness, the other optimized for performance. Take this class for example:

  public class DataRequirementSort : Element {
    [JsonProperty("direction")]
    public string Direction { get; set; }
    [JsonProperty("_direction")]
    public Element _Direction { get; set; }
    [JsonProperty("path")]
    public string Path { get; set; }
    [JsonProperty("_path")]
    public Element _Path { get; set; }
  }

For developer-friendliness, it might look like this:

public class DataRequirementSort : Element 
{
    [JsonProperty("direction")]
    public Element<string> Direction { get; set; }
    [JsonProperty("path")]
    public Element<string> Path { get; set; }
}

internal class Element<T>
{
    public Element<string> Id { get; set; }
    public Extension[] Extensions { get; set; }
    public T Value { get; set; }
}

For server-side performance, it might look more like this, given that extensions on primitive types are probably exceedingly rare:

internal class DataRequirementSort : Element
{
    private Dictionary<int, Element> _primitiveElements;

    [JsonProperty("direction")]
    public string Direction { get; set; }
    
    [JsonProperty("_direction")]
    public Element _Direction
    {
        get => GetPrimitiveElement(0);
        set => SetPrimitiveElement(0, value);
    }
    
    [JsonProperty("path")]
    public string Path { get; set; }
    
    [JsonProperty("_path")]
    public Element _Path
    {
        get => GetPrimitiveElement(0);
        set => SetPrimitiveElement(0, value);
    }

    private Element GetPrimitiveElement(int id)
    {
        Element value = null;
        _primitiveElements?.TryGetValue(id, out value);
        return value;
    }

    private void SetPrimitiveElement(int id, Element value)
    {
        (_primitiveElements ?? (_primitiveElements = new Dictionary<int, Element>()))[id] = value;
    }
}

Originally posted by @johnstairs in #2 (comment)

My company is looking into using more and more custom FHIR resources, So I'm looking to develop a source generator to auto generate the resource code based on the structure definition. My idea is that it will look for files with a name that ends with .StructureDefinition.json and then generate code from those files.

I tried my hand in this but I think this could work.

I'm currently working on this branch:
https://github.com/almostchristian/fhir-codegen/tree/feature/sourcegen

Some issues that I'm looking to resolving:

• change namespace to be the same as the <assemblyname>.<folder.path> instead of the constant namespace. This should be a simple change. supported in latest change
• nested enum and generic Code<T> property generation, I'm still trying to figure out how to get this to work. supported in latest change

Exception at FhirSlicing.cs:70

Example of offending slicing:

"slicing": {
                    "discriminator": [
                        {
                            "type": "type",
                            "path": "$this.resolve()"
                        },
                        {
                            "type": "profile",
                            "path": "$this.resolve()"
                        }
                    ],
                    "rules": "open"
                }

After building on an M1 macOS machine successfully with a brew install dotnet v6.0.103, running the project results in the following error attempting to link in an x64 library. How do you build and run on machines using Apple's current M1 chips? The link printed by the runtime leads me to believe this should work somehow.

`
➜ fhir-codegen git:(main) dotnet run --project src/fhir-codegen-cli
Failed to load Q, error: dlopen(/opt/homebrew/Cellar/dotnet/6.0.103/libexec/host/fxr/6.0.3/libhostfxr.dylib, 0x0001): tried: '/opt/homebrew/Cellar/dotnet/6.0.103/libexec/host/fxr/6.0.3/libhostfxr.dylib' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64'))
The library libhostfxr.dylib was found, but loading it from /opt/homebrew/Cellar/dotnet/6.0.103/libexec/host/fxr/6.0.3/libhostfxr.dylib failed

• Installing .NET Core prerequisites might help resolve this problem.
  https://go.microsoft.com/fwlink/?linkid=2063366
  `

I am currently using the fhir-codegen to generate the vallue sets frome the FrCore Project.

I would like to know if it would be possible to generate the classes that represents the extension. From what i've read, i've understood that for the moment it is not possible to generate classes for the profiles according to their structure definition, but what about the extensions and the slices (for example the Identifiers for Patient ) ? It would make me avoid to write the extensions classes manually and generate them from the tool..

Thank's for reading and for your attention !

Can we split them up so that we can use these serializers when just working with the "common" classes in SDK.

We "just" have to fix the Extension.value[x] problem. Maybe with configurable factories that know how to handle instantiating types, even if common doesn't know about it?

(Let me know if you'd prefer I split this issue up more!)

We've been doing some prototypes based on the TypeScript R4 .ts file, and have run into a few challenges I think could benefit from some general refactoring.

First is the lack of a .d.ts file. We tried simply renaming the .ts, but it doesn't work since .d.ts declarations can't have initializers. (Might be other issues.) Second is general performance. The file is > 400,000 lines long, causing editors to choke. Even when cherrypicking specific imports, VS Code, at least, seems to not handle the single file approach well. Third is the general approach to generics covered in #70 that makes it very challenging to implement concrete class definitions for FHIR profiles imposing type and cardinality constraints.

If the direction is to also provide a set of class definitions like the C# version -- which we would love and appreciate! -- I would suggest breaking up the output file substantially, implement some uses of generics (at least for frequently used types), and to add a corresponding concrete class hierarchy with essentially the same deep copy and serialization-type functions as C# outputs.

All this would really make it directly usable for SMART and general FHIR developers. At the moment we can't really base our class definitions off of it because we're trying to enforce profile constraints at a more granular, static level.

Improve performance by using System.Text.Json instead of Newtonsoft.Json for reading and writing JSON files.

For choice properties, did you consider encapsulating the value in a union-like type and having just one top-level property? So an example might look like this:

Patient
  Deceased
    Boolean
    DateTime

instead of

Patient
  DeceasedBoolean
  DeceasedDateTime

Originally posted by @johnstairs in #2 (comment)

When I do an export (eg CSharp2) from any package it does say that it successfully completed, but I can't find the file anywhere:

As reported here https://chat.fhir.org/#narrow/stream/179171-dotnet/topic/Microsoft.20fhir-codegen-cli/near/383005716

Would it be possible to make the TypeScript version of Reference generic? It is very difficult to enforce constraints without this, and TypeScript supports complex "A | B | C | D"-style generics.

We are trying to strictly enforce all manner of constraints in our information models for both IG publishing and software development purposes, and such refinements would be very welcome!

I would suggest IList instead of T[] for collection properties for versatility. Also I presume it would be more efficient for deserialization since the value does not have to be "trimmed" to an array

Originally posted by @johnstairs in #2 (comment)