- Luca Casonato
- Bret Little
- Chengzhong Wu
- Colin Ihrig
- Dylan Thacker-Smith
- Filip Skokan
- James M Snell
- Steve Faulkner
- Specify
globalThisextendingEventTargetin a seperate document. James will look into it. - Chairs to set up a workstream for environment metadata. Bret will champion.
- Chairs to set up a meeting for the fetch workstream with the relevant parties.
- Chairs to set up a workstream for performance measurements. Chengzhong will champion.
- Introductions
- Updates from workstreams
- Minimum Common API:
- Should
globalThisextendEventTarget?- It does in Deno and CFW.
- James says it is questionable whether this would be implemented in Node.
- Could this be added as a normative optional feature?
- Filip argues that the common minimum API should probably not have optional features, because it wouldn't be common minimal then.
- ACTION ITEM: Specify this in a seperate document. James will look into it.
- Should user agent be a standard format?
- OpenTelemetry uses the user agent as an opaque string now, but would like more structured info.
- Probably not a standard format, as we don't want to encourage parsing.
- We should likely use "user agent data" system that Chromium has implemented. This is more structured info than parsing a user agent string.
- User agent data can also contain "stacked" vendor lists, like
["Chromium", "Chrome"]which may translate over to server side runtimes well.
- Injection of environment variables / environment metadata:
- It would be nice to have a consistent way to read environment metadata from the runtime.
- In Deno and Node this is achieved using environment variables.
- In CFW "secrets" and environment variables are bundled together with KV and durable object bindings in a single context object.
- Luca proposes this could be a string -> string map on
navigator(navigator.envfor example). - ACTION ITEM: Chairs to set up a workstream
- Bret to champion this workstream.
- Should
- WebCrypto Streams:
- Reached out to Mozilla. They consider the proposal non harmful, but have
some concerns about specific algorithms in
EncryptionStream.
- Reached out to Mozilla. They consider the proposal non harmful, but have
some concerns about specific algorithms in
- Fetch:
- No progress.
- ACTION ITEM: Chairs to set up a meeting with relevant parties.
- Minimum Common API:
- Is Node going to join the group?
- No progress on this.
- ACTION ITEM: James will push this forward.
Performanceglobal- Are runtimes implementing this, and what are concerns that need to be addressed?
- Garbage collection of marks and measures is an issue. Would it help to have a global limit that the runtime enforces?
- ACTION ITEM: Chairs to set up a workstream
- Chengzhong to champion this workstream.