[eli-r-ph]

💡 Motivation and Context

The capture v1 endpoint parses each event uuid leniently (uppercase, no hyphens, braced, urn:uuid:), but keys the 200 results map by the canonical lowercase hyphenated form (response.rs).

makeUUID validated a caller-supplied uuid but returned it unchanged, and partitionResults looks results up by the exact string sent. For a non-canonical uuid the lookup missed, so the event was treated as accepted: a retry was never retried, a drop never reached Callback.Failure, and no success callback fired either.

• makeUUID now returns the canonical form of a valid caller-supplied uuid.
• prepareForSend canonicalizes the correlation key, because BeforeSend runs after makeUUID and may set any uuid form.

posthog-rs is not affected: its event uuid is a typed Uuid, which always serializes canonically.

💚 How did you test it?

• New TestSendCorrelatesNonCanonicalUuid sends uppercase, unhyphenated, braced and urn:uuid: uuids to a test server that keys drop results canonically, and asserts the failure callback fires. It fails without the fix.
• TestGeneratedUUIDsAreV7 asserts caller-supplied uuids are canonicalized.
• go test ./... passes except TestOpenMachOUniversal, which also fails on unmodified v1 on my machine (macOS debug-image fixture).

📝 Checklist

• I reviewed the submitted code.
• I added tests to verify the changes.
• I updated the docs if needed.
• No breaking change or entry added to the changelog.

If releasing new changes

• Ran pnpm changeset to generate a changeset file

🤖 Agent context

Autonomy: Human-driven (agent-assisted)

Written with Cursor (Claude Opus) under the direction of the assignee. Found while aligning posthog-python's capture v1 result handling with this branch. "Missing from results means accepted" stays as is: with canonical keys, a 2xx always has an entry per event, so that branch only matters if the server breaks its contract.

[eli-r-ph]

Quick parity fix for this compared to other core backend SDKs I'm prepping for v1 release

[dustinbyrne]

The UUID correlation fix looks correct. The malformed-wrapper fallback note below is non-blocking.

AI-assisted review.