Track upstream: Vale lint time is superlinear in single-block size, and no output is emitted until the run finishes

[thecodedrift]

Resolved upstream in Vale 3.21.0, which 0.11.2 ships (2026-09-20). Reproduction repo with the timing data: https://github.com/thecodedrift/reproduction_vale_blocks. The 128KB guard 0.11.2 added against the old behavior comes out in the next patch: #351.

[thecodedrift]

Direction settled: this issue is the one we are betting on. #324 (batching) is closed in its favour.

The position we are standing on until upstream moves:

• VALE_MAX_FILE_BYTES (fix(cli): skip an oversized file instead of letting Vale time out on it #323) stays as the guard. It avoids the spend rather than spending and discarding it, and it is the only check cheap enough to run before spawning Vale.
• We build no workaround for the all-or-nothing output. Batching would have made a timeout cost one batch instead of the whole run, and its entire value disappears if finding 2 here is fixed. Not worth carrying a workaround whose lifetime is bounded by a bug report we have already filed.
• The size cap is knowingly a proxy for the wrong quantity, per finding 1: cost tracks single-block size, not file size. It over-fires on a well-formed 3 MB document (~4s, passes a 128 KB cap) and under-fires on a 512 KB single-block file (~2s, may sit beneath it). Accepted, because measuring block structure means reading every candidate, which costs more than the guard saves.

Either upstream fix changes what we do here:

• Incremental output (finding 2) removes the reason batching ever existed, and makes a caller-side timeout able to preserve findings.
• Linear block cost (finding 1) lets VALE_MAX_FILE_BYTES rise substantially or disappear.

A "won't fix" upstream is the trigger to reopen #324, not a reason to start it now.

Reproduction repo, with the vale binary pinned via npm and npm run bug:reproduce: https://github.com/thecodedrift/reproduction_vale_blocks

[thecodedrift]

Verified with @taskless/vale-*@3.21.0-20260915061224: the 3 MB single-block fixture drops from ~81 s to ~230 ms and now matches the many-block file (~270 ms), with identical alert counts. Scaling a single block to 25 MB gives ~1.8 s, linear in size. Vale still emits nothing until the whole run finishes, but with no slow block for a timeout to hit that is no longer causing lost output; treating this as fixed for Vale ≥ 3.21 and archiving the repro repo.