Re: HTTP/2 Header Encoding Status Update

In message <5134B678.2010308@cisco.com>, Eliot Lear writes:

>> At present there are no relevant time formats which are leap-second safe.
>
>>From a *format* perspective, at least ISO-8601 and RFC-5322 (Message
>Format) are examples where leap-seconds are supported.

Both are standards for textual (aka: human readable) representation
of timestamps and involve a lot of text-processing to perform the
for HTTP usage necessary before/after comparisons on.

If HTTP/2 has any pretentions of being a high-performance protocol,
it must define an arithmetic time-representation, which allows
simple and cheap comparisons and the arithmetic operations necessary.

>When talking about seconds
>from an epoch, it seems to me that if the second occurred it should be
>counted, but I would suspect there already is a standard there as well,
>and we should follow it.  What would Linux do? ;-)

UNIX, POSIX, Linux and Windows all pretends that leapseconds don't exist.

-- 
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, 4 March 2013 15:08:37 UTC