Patrick McManus

https://github.com/quicwg/base-drafts/issues/1257 clarifies that this is 3 out per 1 in

I think this means the library can have a h1 streamed interface - perhaps with the a f(x) to delimit headers instead of 2xcrlf and native qpack support.. then the...

* change ip address (both sides) * validation * new_connection_id

the library now provides more stream events if you do a recv() while handling the previous one (and there is more data buffered). this can help.

i don't think actually wants an api for setting ciphers - this is something it should simply do correctly as a design choice. but its a design choice and ymmv...

feel free to make the change if you like.

this can be done quietly for the user though, right? through init_once and ref counting? On Mon, Jan 23, 2017 at 11:59 PM, Daniel Stenberg wrote: > (Discussion forked off...

thread lib seems like an ok requirement to me in 2017.. if you really don't want it you can build an atomic spinlock for arm and x86 easily enough for...

the problem with explicit global init is that it doesn't necessarily embed well into other libraries.. the library can't really tell if it has been init'd already/concurrently what not.. On...

cc @mnot - I'm a bit confused on the context - is this saying that 2 different uris with 206 responses should be stitched together just because they both had...