@ddbeck When you get a chance, could you look at this. Mostly I am confirming that this is reasonable to have in the BCD (at all). I can then perhaps do some experimentation to determine versions it might appear.

@hamishwillee when you say that anything that supports WebSockets should support this, are you saying that the browser should send an Upgrade header as part of creating session? My thinking here is this: if we can show that a browser will use this header of its own accord (without a web developer composing HTTP requests themselves—at which point they could use whatever headers they please), then I think we could have some compat data for it here. Otherwise, I think we should omit it.

when you say that anything that supports WebSockets should support this, are you saying that the browser should send an Upgrade header as part of creating session?

Yes, that is my understanding. Any system that supports web sockets via HTTP1.1 will create a normal HTTP connection then send UPGRADE to upgrade to a socket protocol. The "How it works" is extensively documented on MDN, and you can see it in the various points in the code on search.

This also how a system with an HTTP1.1 connection might upgrade to HTTP 2.0 (see here).

Anyway, sounds like this is worth having BCD for. Not sure how we'll work out how to verify this, and particular versions. Theoretically we could probably just copy the BCD for websockets, since this is main use case: https://developer.mozilla.org/en-US/docs/Web/API/WebSockets_API

@Elchi3 Do you have any thoughts on how we might confirm support for such a feature? I haven't reviewed much of the HTTP data and wanted to double check with you first.

@Elchi3 Do you have any thoughts on how we might confirm support for such a feature? I haven't reviewed much of the HTTP data and wanted to double check with you first.

I think it is difficult and not worth spending cycles on.

I have more thoughts on this matter, you can find them in this thread from August 2019 (it is still valid): https://discourse.mozilla.org/t/bcd-and-want-digest-digest-http-headers/43340 (basically there are headers where browser support matters and also where it doesn't).

Thanks @Elchi3 . I've accepted the proposed changes.

I think this is a case where it is useful to state support, and doing that with "true" is better than nothing (i.e. option "3" in linked doc). More generally, if it isn't clear what support means OR we can't verify support because it depends on the server I lean towards option 4 or omitting the section altogether.

@ddbeck Ready to merge, or will do "option 4" if you just close this (indicating you think it isn't of value)