Consider renaming or replacing http3only?
ekinnear opened this issue · 1 comments
What is the issue with the Fetch Standard?
In the context of WebTransport, the use of the term http3only in obtain a connection came up in discussion of w3c/webtransport#561.
We've been trying to adhere to more transport agnostic principles, allowing developers to express their needs based on properties of the transport, rather than hardcoding a particular protocol or a version of a protocol.
In that context, the reason for requiring http3only was determined to instead represent the desire to obtain unreliable transport that can eliminate head-of-line blocking, perhaps we should use some spelling of the term requireUnreliableTransport.
Filing an issue to discuss if we want to transition to something more transport agnostic in fetch as well when obtaining connections. What happens when we have HTTP/4? HTTP/3.1? What if HTTP/3.1 doesn't provide the same underlying transport properties? It seems like enshrining the protocol version itself as the name of the field leaves us in an undesirable position.