wasmtime / PR #14448 cli: reject zero for the --http-outg... · git-wasmtime

Stream: git-wasmtime

Topic: wasmtime / PR #14448 cli: reject zero for the --http-outg...

Wasmtime GitHub notifications bot (Sep 29 2026 at 17:41):

xia-chao opened PR #14448 from xia-chao:reject-zero-http-outgoing-body-options to bytecodealliance:main:

Both options must be non-zero at the implementation level. Passing zero used to misbehave: buffer-chunks=0 panicked a worker under serve and the whole process under run, and chunk-size=0 returned an empty body with a 200. They are now Option<NonZeroUsize>, like gc_zeal_alloc_counter, so valid configurations are unaffected.

Wasmtime GitHub notifications bot (Sep 29 2026 at 17:41):

xia-chao requested wasmtime-core-reviewers for a review on PR #14448.

Wasmtime GitHub notifications bot (Sep 29 2026 at 17:41):

xia-chao requested cfallin for a review on PR #14448.

Wasmtime GitHub notifications bot (Sep 29 2026 at 17:57):

xia-chao edited PR #14448:

Both options must be non-zero at the implementation level. Passing zero used to misbehave: buffer-chunks=0 panicked a worker under serve and the whole process under run, and chunk-size=0 returned an empty body with a 200.

Wasmtime GitHub notifications bot (Sep 29 2026 at 18:00):

xia-chao edited PR #14448:

Both options must be non-zero at the implementation level. Passing zero used to misbehave: buffer-chunks=0 panicked a worker under serve and the whole process under run, and chunk-size=0 returned an empty body with a 200.
For more information, please refer to diff.

Wasmtime GitHub notifications bot (Sep 29 2026 at 18:18):

:speech_balloon: cfallin created PR review comment:

We can probably rely just on the cli-test to assert the behavior here -- we don't have unit tests for any other option-specific semantic validation here and it seems sort of repetitive.