[release/6.0] Fix custom JsonConverterFactories not working with nested type/property declarations.

[github-actions]

Backport of #62595 to release/6.0

/cc @eiriktsarpalis

Customer Impact

Customer reported issues in #62079 and #61860. Fixes a bug in which System.Text.Json will generate invalid code in cases where transitive dependencies in the type graph of a serializable type specify custom JsonConverterFactories via the JsonConverter attribute.

Testing

Added regression tests validating the impacted use cases.

Risk

Low to moderate. Makes a minor adjustment in the product code.

[ghost]

Tagging subscribers to this area: @dotnet/area-system-text-json
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Backport of #62595 to release/6.0

/cc @eiriktsarpalis

Customer Impact

Testing

Risk

Author:	github-actions[bot]
Assignees:	-
Labels:	area-System.Text.Json
Milestone:	-

[danmoseley]

@ericstj could you please handle getting approval for this? by email or on Thurs.

[ericstj]

nit: at first it wasn't clear to me that these were tied to the _currentContext and that _currentContext would actually change as the emitter did its work. It might make sense to associate this mutable state to a single object or name it in a way that makes it clear how it applies and when it changes.

[ericstj]

For example as members of ContextGenerationSpec

[steveharter]

LGTM; allows custom converters for types not adorned with [JsonSerializable]

[ericstj]

Approved over email.

[danmoseley]

is this ready to merge?

[danmoseley]

test failures will be fixed by #63526. this can merge if it's ready.

[danmoseley]

Let's re-run CI to verify those are fixed, since it's Sunday..

[ericstj]

@safern for some reason we're seeing PackageValidation fail again like it was here: #59908. I'm not sure why that might have just started happening. Maybe because this is a rather old PR and is running old pipelines?

[eiriktsarpalis]

Maybe it just needs rebasing?

[safern]

@safern for some reason we're seeing PackageValidation fail again like it was here: #59908. I'm not sure why that might have just started happening. Maybe because this is a rather old PR and is running old pipelines?

I've seen it in multiple PRs already, I think I need to service the fix to 6.0 on the SDK and update the package. I'll do that for 6.0.3 as we don't have time for 6.0.2 anymore.