W3C home > Mailing lists > Public > public-web-perf@w3.org > September 2013

Re: [ResourceTiming] Wire Protocol

From: Nottingham, Mark <mnotting@akamai.com>
Date: Thu, 5 Sep 2013 19:21:41 -0500
To: "McCall, Mike" <mmccall@akamai.com>
CC: public-web-perf <public-web-perf@w3.org>, Jatinder Mann <jmann@microsoft.com>
Message-ID: <AE02FFD7-F421-42A0-869B-FF470D26E152@akamai.com>
That's why I think Resource Timing should just specify a container for this information, not the range of possible values; leave that up to the registry (for standard ones) and implementations (for draft / experimental ones).

Cheers,


On 06/09/2013, at 12:23 AM, "McCall, Mike" <mmccall@akamai.com> wrote:

> In yesterday's meeting minutes, a few people mentioned that the only
> protocols enumerated by resource timing be strictly HTTP-based.  While I
> can agree in principle, the web moves much faster than standards bodies.
> For example, SPDY first came to light in 2009; it's taken almost four
> years for it to get close to being called "HTTP".
> 
> What would happen if a UA supported a proprietary protocol, and fetched a
> resource via that protocol from the server?  Would Resource Timing ignore
> it because it doesn't fall into the enumerated HTTP-only list, leave it
> blank, or worse, mislabel it as HTTP?  I think that if a resource was
> fetched by a User Agent via HTTP/SPDY/QUIC/etc., it should be labeled as
> such.
> 
> Mike
> 
> On 9/4/13 12:33 PM, "Jatinder Mann" <jmann@microsoft.com> wrote:
> 
>> Seeing that all major browser vendors now have some SPDY implementation,
>> it seems reasonable that we should consider adding wire protocol
>> information to Resource Timing L2 spec. Whether this new attribute
>> returns HTTP2, SPDY1/2/3, or something else should be discussed.
>> 
>> Thanks,
>> Jatinder
>> 
>> -----Original Message-----
>> From: Nottingham, Mark [mailto:mnotting@akamai.com]
>> Sent: Monday, September 2, 2013 9:42 PM
>> To: public-web-perf
>> Subject: [ResourceTiming] Wire Protocol
>> 
>> Hi,
>> 
>> Previously, there have been requests [1] to add information to
>> ResourceTiming about the wire protocol used in fetches, in particular for
>> SPDY.
>> 
>> There was a bit of discussion, but it seemed to stop here [2]:
>> 
>>> I think it's too early to add spdy/http2 vs. http given that the
>>> protocol is still in works.
>> 
>> 
>> HTTP/2.0 [3] is now seeing early implementation [4]. Since the focus of
>> HTTP/2.0 is performance, having metrics about how it is performing is
>> critical to evaluate its effectiveness and aid its adoption.
>> 
>> Is it really the intent of the WG to wait until the protocol is no longer
>> "still in the works" before adding a metrics for it?
>> 
>> I'm happy to make a detailed proposal if that will help unblock this...
>> 
>> Thanks,
>> 
>> 1. http://lists.w3.org/Archives/Public/public-web-perf/2013Jan/0022.html
>> 2. http://lists.w3.org/Archives/Public/public-web-perf/2013Jan/0031.html
>> 3. https://github.com/http2/http2-spec
>> 4. https://github.com/http2/http2-spec/wiki/Implementations
>> 
>> --
>> Mark Nottingham    mnot@akamai.com    http://www.mnot.net/
>> 
>> 
>> 
> 

--
Mark Nottingham    mnot@akamai.com    http://www.mnot.net/
Received on Friday, 6 September 2013 00:22:09 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 18:01:21 UTC