Re: HTTP/2 and NCSA

On 01/07/2014 08:33 AM, Mark Nottingham wrote:
> 
> On 7 Jan 2014, at 12:37 am, Poul-Henning Kamp <phk@phk.freebsd.dk>
> wrote:
> 
>> In message <alpine.DEB.2.00.1401061128260.3079@wnl.j3.bet>, Yves
>> Lafon writes:
>> 
>>>> One of my users asked and I couldn't answer, so I'm passing it
>>>> on:
>>>> 
>>>> Who will (and when will they) standardize how NCSA log format 
>>>> works with HTTP/2.0, in particular for "new" stuff like PUSH ?
>>>> 
>>>> My cursory research indicates that W3C has documented the NCSA 
>>>> format (somewhat) but it is not clear to me if they actually 
>>>> consider themselves steward of the standardisation ?
>>> 
>>> The document is from... 1996 [1], so the space is up for grab.
>>> 
>>> [1] http://www.w3.org/TR/WD-logfile.html
>> 
>> Yes, that's how I derived that W3C didn't seem to be very
>> interested.
>> 
>> Are we ?
> 
> Hmm. It’s not really a standard in the sense that IETF does them, and
> I don’t know of any other log formats that have been standardised in
> that sense.
> 
> That said, if an implementation wanted to publish an Informational
> RFC documenting their HTTP2-friendly log format, that’s entirely
> doable, and I’m sure they’d get some help if they asked.

I've no opinion of whether a log format should or should
not be "part" of the HTTP/2.0 work. But the IETF has done
similar stuff - SIPCLF [1] was a WG that set out to define
a log file format for SIP that's based on the Apache log
file format.

Unfortunately they never thought about privacy when doing
that, and by the time I heard about it, it was too late.
E.g. there's no well-defined way to obsfuscate IP addresses
so they can be correlated for a while only. If someone were
to try do such work for HTTP/2.0 I'd hope that they would
seriously consider privacy issues.

S.

[1] http://tools.ietf.org/wg/sipclf/

> 
> Cheers,
> 
> 
> 
> Mark Nottingham   http://www.mnot.net/
> 
> 
> 
> 
> 
>

Received on Tuesday, 7 January 2014 13:58:41 UTC