Add top-level type any as semantic version of untyped

[Strech]

What does this PR do?

Add a Datadog top-level semantic type any.

Motivation:

The type any is going to represent a scenario when the narrowing is impossible to do or when it doesn't make sense to do. And type untyped should represent an actual untyped type (no time, not done, etc)

Change log entry

No.

Additional Notes:

We can get more value out of untyped checks once we start moving to any in case when it's really any type.

How to test the change?

CI

[lloeki]

Very cool.

AIUI both are logically equivalent type-wise, they are semantically not:

• untyped: "I did not have time to define the proper type for this, so I accept everything and the name serves as a marker to revisit later."
• any: "I did have time to thoroughly think about what this type should be and it is definitely intended to accept every possible type, so I am communicating that this is intendend as the proper type here."

It's a bit like:

void, boolish, or top?

They are all equivalent for the type system; they are all top type.

void tells developers a hint that the value should not be used. boolish implies the value is used as a truth value. top is anything else.

For context any was a thing but was renamed untyped:

sig/datadog/open_feature/exposures/worker.rbs:48:23: [error] Cannot find type `any`
│ Diagnostic ID: RBS::UnknownTypeName
│
└         def perform: (*any) -> void

Maybe we can make a case upstream to have both?

[lloeki]

LGTM

[pr-commenter]

Benchmarks

Benchmark execution time: 2025-11-18 10:13:15

Comparing candidate commit f4b627b in PR branch no-ticket-add-top-level-any-rbs-type with baseline commit f33bab7 in branch master.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 44 metrics, 2 unstable metrics.

[y9v]

I definitely find the proposed change useful