Re: [quicwg/base-drafts] PUSH_ID as a frame (#2526)

[Luca Niccolini <[e-mail]>]

I was not trying to solve the same problem as #2527 here, but I was mentioning the fact that I believe framing would slightly help in a subset of cases. 
Take a server with a buggy push implementation, rather than a malicious one. Something that doesn't send the PUSH_ID at all. On the client you'll start seeing a flow of data and you can't do anything better than parsing the first few bytes of that as a PUSH_ID, and basically see random IDs. With Framing, if a server forgets to send the PUSH_ID, it is clear at the receiver that it is a protocol error.
That being said there are many other classes of bugs that may trigger the same behavior even with frames, so this is minor. 


> frame processing should only take place once the PUSH_PROMISE is received. 
@LPardue now, that seems like an implementation dependent claim :)
 
as #2527 says (and I like it) a client should buffer any data after receiving the PUSH_ID. and that is sufficient. no mention of frames there.


-- 
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/issues/2526#issuecomment-473915586