Tankcorn/preview observability streaming

[Ankcorn]

This PR adds preview metadata to the streaming tail worker onset event.

It follows the same pattern already used for scriptVersion: the metadata is attached to Onset::WorkerInfo, serialized through the onset RPC
payload, and exposed in the JS event delivered to streaming tail workers.

What changed

Streaming tail onset events now include:

preview: {
     id,
     slug,
     name
}

[anonrig]

Suggested change

	JSG_MEMORY_INFO(Preview) {
	tracker.trackField("id", id);
	tracker.trackField("slug", slug);
	tracker.trackField("name", name);
	}

Not needed

[Ankcorn]

When I removed these I think it broke the build

Error while processing /home/runner/.cache/bazel/_bazel_runner/be54bd8ac890d1817c772d91eb8f6cb4/sandbox/linux-sandbox/7561/execroot/_main/src/workerd/api/trace.c++.
bazel-out/k8-fastbuild/bin/src/workerd/jsg/_virtual_includes/memory-tracker/workerd/jsg/memory.h:234:8: note: candidate template ignored: could not match 'const kj::Table<T, Indexes...>' against 'const workerd::api::TracePreviewInfo'
234 | void trackField(kj::StringPtr edgeName,
| ^

I have kept them in for now but if there is something I should be doing instead please lmk

[codspeed-hq]

Merging this PR will improve performance by 15.28%

⚡ 1 improved benchmark
✅ 69 untouched benchmarks
⏩ 129 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	Encode_ASCII_32[TextEncoder][0/0/32]	3 ms	2.6 ms	+15.28%

---

_{Comparing Ankcorn:tankcorn/preview-observability-streaming (a1dfdeb) with main (866c9c7)}

Footnotes

1. 129 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[danlapid]

/bonk review

[ask-bonk]

Clean, well-structured change. The preview metadata is threaded through all the right layers (capnp schema, C++ serialization/deserialization, clone, streaming tail JS serialization) following the same pattern as scriptVersion. Tests cover the capnp round-trip and clone for the Onset path.

One minor nit (not blocking): The comment on TracePreview in src/workerd/io/trace.h:434 says "Describes a buffered tail worker request" which doesn't accurately describe what it represents. Since it holds preview deployment metadata (id, slug, name), something like "Describes preview deployment metadata associated with a worker" would be clearer. This predates your PR but worth fixing while you're touching this area.

[ask-bonk]

Review submitted. Here's a summary of my findings:

Verdict: Approved

The PR cleanly adds preview deployment metadata (id, slug, name) to the streaming tail worker onset event. It touches all the right layers:

1. worker-interface.capnp - Adds preview @10 :TracePreviewInfo to the Onset struct with the correct next ordinal
2. trace.h - Adds kj::Maybe<TracePreview> preview to Onset::WorkerInfo
3. trace.c++ - Adds deserialization (getPreviewFromReader), serialization (copyTo), and clone() support
4. tracer.c++ - Copies preview from Trace into the WorkerInfo when constructing the onset event
5. trace-stream.c++ - Serializes preview to JS object with id, slug, name fields (the ToJs path for streaming tail workers)
6. trace-test.c++ - Tests capnp round-trip and clone for the Onset path

The implementation follows the exact same pattern as scriptVersion, which is the right approach. One minor nit about an inaccurate pre-existing comment on TracePreview.

github run