Re: [quicwg/base-drafts] Length-prefixes and flow control (#1432)

[Mike Bishop <[e-mail]>]

IIRC, HPACK EOS is so long because you never actually send all of it – if the Huffman-encoded string doesn’t end on a byte boundary (and it probably doesn’t), you put as many bits of the EOS symbol as needed to reach the byte boundary.  The whole thing should never appear on the wire; at most seven bits of it.  Of course, if you’re depending on recognizing it as it goes by, that requires parsing each symbol as it passes; the length prefix makes it much easier.

From: Dmitri Tikhonov <[e-mail]>
Sent: Monday, June 11, 2018 5:43 PM
To: quicwg/base-drafts <[e-mail]>
Cc: Mike Bishop <[e-mail]>; Comment <[e-mail]>
Subject: Re: [quicwg/base-drafts] Length-prefixes and flow control (#1432)


@martinthomson<https://github.com/martinthomson>, mine wasn't a suggestion, but rather something to keep in mind as we weigh different options. That is, there is a way to not use prefix length at all. Your analysis is correct. The biggest downside to me is that the EOS symbol is so long.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub<https://github.com/quicwg/base-drafts/issues/1432#issuecomment-396429945>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AEE2ha2hZGTKuCUGPvAxuog8eYQzTuTCks5t7w6KgaJpZM4UgCd3>.


-- 
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/1432#issuecomment-396458126