[compiler] Fix false negative: independently memoize values used as method call receivers

[Felipeness]

Summary

Fixes #35902

When a method like .map() is called on a value with an unknown type (e.g. the return value of an unknown function like expensiveProcessing(data)), the compiler had no function signature available. This caused the fallback in InferMutationAliasingEffects to apply MutateTransitiveConditionally on the receiver, over-extending its mutable range and merging it into the same reactive scope as the method call result.

Before (bug): expensiveProcessing(data) re-executes when onClick changes:

if ($[0] !== data || $[1] !== onClick) {
  const processedData = expensiveProcessing(data); // unnecessary!
  t1 = processedData.map(t2);
}

After (fix): expensiveProcessing(data) has its own scope depending only on data:

if ($[0] !== data) {
  t1 = expensiveProcessing(data);
}
const processedData = t1;
if ($[2] !== onClick || $[3] !== processedData) {
  t2 = <ul>{processedData.map(...)}</ul>;
}

Root cause

In getPropertyType(), when the receiver has an unresolved type (Type kind), property lookups returned null — meaning no function signature was available for the method call. Without a signature, InferMutationAliasingEffects falls back to MutateTransitiveConditionally on all operands including the receiver, which extends its mutable range through the method call. This causes InferReactiveScopeVariables (union-find) to group the receiver's creation and the method call into the same reactive scope.

Fix

Added a fallback in getPropertyType: when the receiver has an unresolved type, try to resolve the property from the BuiltInArray shape — but only if the method also exists in BuiltInMixedReadonly (the safe, non-mutating subset). This follows the same reasoning already documented in ObjectShape.ts for BuiltInMixedReadonlyId:

"if .map() is called on an unknown type, the value must be array-like"

We use the BuiltInArray shape (not BuiltInMixedReadonly) because it has aliasing signatures that correctly model data flow without over-extending mutable ranges. The MixedReadonly guard ensures that mutating methods like .push() and .pop() — which only exist in BuiltInArray — are NOT resolved through this fallback, preserving the conservative behavior for those cases.

Test plan

• Added fixture bug-independent-memoization-for-unknown-receiver-method-call demonstrating the fix
• Verified 1090 existing fixtures produce identical output (zero regressions)
• Manually tested: .push() (mutating, not in MixedReadonly) still uses conservative fallback
• Manually tested: .filter(), .map() on unknown types now correctly separate scopes
• Manually tested: custom methods (not in any shape) still use conservative fallback

[meta-cla]

Hi @Felipeness!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks!

[meta-cla]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks!

[josephsavona]

This isn’t safe. I believe in commented on a similar PR to this one, but tldr is that if we don’t know the type, then we don’t know that it’s map/filter/etc method is actually non-mutating.

[josephsavona]

Closing since there is no variant of this that works safely. The long-term solution that we’re exploring is using information from type systems (Flow, hopefully also TS) in order to have more information to guide compilation.