Re: [quicwg/base-drafts] Allow the HTTP/3 DATA frame to extend to the end of the stream (#2098)

[Mike Bishop <[e-mail]>]

I'm with @dtikhonov on this; it's not wholly implausible to have an extension frame where receipt of the frame is itself the message and no payload is necessary.  That said, it's also a con of making this a general property of frames, since a frame with an empty payload which is not the last frame of the stream becomes impossible.

But slight caveat to @dtikhonov's earlier comment:  I wear two "hats" on this document, and as editor, my job is to make the document reflect my understanding of the consensus.  My opinion as a contributor shouldn't count any more than anyone else's because of that other hat.  (And my apologies if I cross wires -- please call that out if you think I am.)

So by my count, we're about evenly split on whether this should be a general property of all frames or specific to DATA, with most of us in both directions stating that it's a slight preference or that the other is acceptable.  There's a technical advantage to each (simpler processing if DATA-only, ability to use for no-body requests if HEADERS too).

That makes me inclined to leave it to the author of the PR, unless someone objects violently.  @RyanAtGoogle, let me know when you've incorporated any feedback you plan to.  Anyone who plans to object violently, speak now while I'm at a safe distance.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/pull/2098#issuecomment-444592338