fix(rpc): honor deprecated attribute keys in trace item attribute values

[xurui-c]

https://linear.app/getsentry/issue/EAP-459/attributevaluesrequest-returns-empty-results-for-attributes-with

Fix the Trace Item Attribute Values RPC when clients request the canonical attribute name but data is stored only under a deprecated or alias key (e.g. db.system.name vs db.system).

attribute_key_to_expression already coalesces those names via ATTRIBUTES_TO_COALESCE, but the endpoint only checked has(attributes_string, <request key>), so inner rows that held values exclusively on the legacy key never matched. We now OR existence checks across the canonical name and any coalesce targets, and run add_existence_check_to_subscriptable_references on the inner query so behavior stays aligned with other EAP RPC paths.

Adds an integration test that writes db.system and requests db.system.name, asserting the returned distinct values.

Refs:

[Copilot]

Pull request overview

Fixes the Trace Item Attribute Values v1 RPC to return values when clients request a canonical attribute name but events only contain the deprecated/alias key, aligning the endpoint’s filtering behavior with attribute_key_to_expression coalescing.

Changes:

• Update the attribute existence filter to OR across the requested key and any coalesce/deprecated targets.
• Apply add_existence_check_to_subscriptable_references to the inner query to keep missing-key behavior consistent with other EAP RPC queries.
• Add an integration test that writes values under db.system and queries db.system.name.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
snuba/web/rpc/v1/trace_item_attribute_values.py	Extends attribute existence checks to cover coalesced keys and adds subscriptable existence wrapping to the inner query.
tests/web/rpc/v1/test_trace_item_attribute_values_v1.py	Adds coverage for canonical-vs-deprecated attribute key lookups returning distinct values.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

ATTRIBUTES_TO_COALESCE values are set[str], so iterating them makes the generated OR condition order nondeterministic across runs (and across Python hash seeds). This can make query serialization/caching/metrics noisier and can complicate debugging. Consider iterating a sorted list of alternates (e.g. for alt in sorted(...)) to keep the query AST stable.

Suggested change

	for alt in ATTRIBUTES_TO_COALESCE.get(request_key.name, ()):
	for alt in sorted(ATTRIBUTES_TO_COALESCE.get(request_key.name, ())):