Guy Bedford
Guy Bedford
We could support the specified `cache` option for fetch with the following: * `'default'`: This is our default so we exactly match the spec here * `'no-store'`: This is the...
We currently implement this as throwing on invalid UTF8 bytes, but per https://encoding.spec.whatwg.org/#utf-8-decode we should handle lossy string conversion.
We still don't support the generation integer in KV Store lookup and insert APIs, due to the bug fix that is still releasing in Viceroy for this. When that goes...
In the WPT file tests/wpt-harness/expectations/fetch/data-urls/base64.any.js.json, we have failures which are not present on the StarlingMonkey build. We need to track this down further.
Currently we assign dynamic backend names based on the host name only. This results in backends being shared incorrectly when they have different configurations. We should either implement a full...
This support seems to be missing - https://github.com/fastly/js-compute-runtime/blob/e3aa554f31c92e48be30e72c9a4fa45dd135fd79/integration-tests/js-compute/fixtures/app/src/cache-core.js#L1951. Also like KV we seem not so support streams - https://github.com/fastly/js-compute-runtime/blob/e3aa554f31c92e48be30e72c9a4fa45dd135fd79/integration-tests/js-compute/fixtures/app/src/cache-core.js#L496.
When not passing a function to the second argument we should have a better failure mode, in addition when the function does not return a promise we need a better...
There may be a missing error path here when a secret store hostcall like `plaintext` returns an optional none.
This is currently not available in the documentation and provides a useful primitive for determining the HTTP version.
This would be good to ensure for the next major. We can move: * sdkVersion * enableDebugLogging * includeBytes onto the new `fastly:compute`. `setDefaultBackend` can move into `fastly:backend`. Then we...