Re: HTTP 2.0/flag day

In message <CA+9kkMCZPUH1sgWpNhBCN=zeytvjFS8ZufE1KGewFsuh1LJGNQ@mail.gmail.com>
, Ted Hardie writes:

>[...] or do we have a design constraint that all
>use cases currently met by HTTP 1.X must also be met (with the same
>security and performance properties) by 2.0?

It's very expensive to insist on requirements containing phrases
like "all use cases", it typically will cost you a factor of 10
more in effort, relative to writing "all relevant use cases".

I personally would hope that HTTP/2.0 attempts to improve on HTTP/1.1
and thereby replace it, for the vast majority of the use-cases.

I particular I would almost insist on loosing stuff from HTTP/1.1
which is badly thought out (header concatenation), unused (TE: gzip)
or inefficient (CRLF demarkation).

If possible, we should replace it in HTTP/2.0 with stuff which
people want and need, such as a workable session concept, water-tight
cryptographic services, DoS resistance and high performance, realizing
that not every user desires or even want to live with all these features.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

Received on Monday, 26 March 2012 09:47:19 UTC