Re: [quicwg/base-drafts] QPACK improvement: wrap absolute index values (#1644)

[Martin Thomson <[e-mail]>]

I'm also convinced and would like to see a PR.

CeilMul can be 1.  I don't see any point in entertaining anything less (too complicated) and anything more is bananas: you are creating a header block that is blocked for >1 RTT if that arrives.

I think that we should work with 0-based indexing.  HPACK having 1-based indexing was awful and we managed to get rid of that by tweaking the opcodes in QPACK.

Make sure that you include `floor()` in your calculation of MaxEntries.

I like Mike's addendum about 0-RTT.  We will need to be careful about that.  And implementations will need to test that specially.  Of course, for the table to wrap, you need to have evictions in 0-RTT, which won't happen because that takes round trips (at least two, with CeilMul at 1, I believe).  If you have round trips in your 0-RTT, something is badly wrong.

-- 
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/1644#issuecomment-411949968