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

Re: [JSPreflight] An alternate approach

From: McCall, Mike <mmccall@akamai.com>
Date: Mon, 9 Sep 2013 16:37:47 -0400
To: Andy Davies <dajdavies@gmail.com>
CC: "public-web-perf@w3.org" <public-web-perf@w3.org>
Message-ID: <CE53A3F5.1477D%mmccall@akamai.com>
Hi Andy,

Unfortunately, you're probably seeing some of the cruft of the now
year-old proposal.  To some of your questions:

On 9/9/13 3:48 PM, "Andy Davies" <dajdavies@gmail.com> wrote:

>2. Is there a standard for key/value pairs in headers?  X-Timing-Measures
>proposes : ; for assignment and separation whereas  Client-Hints proposes
>= ,

Guypo mentioned using Browser Hints, which is much more well-thought-out
in terms of syntax.  I'd defer to that in terms of what the best delimiter
would be.

> 
>3. I can see a lot of use for this with Nav and User Timing, not so sure
>about Resource Timing due to the potential volume of data (but if it's
>configurable that may not matter)

I agree about Resource Timing, which leads me to...

>4. Will the response always be returned via POST - one advantage of GET
>is people can just have a beacon server that returns a gif or a HTTP 204
>and (in really cheap solutions) the timing data scooped out of the HTTPD
>logfiles.

The POST response was intended specifically to handle Resource Timing.
That being said, if we were to leverage the Beacon API, it offers multiple
verbs for sending back data (GET, POST, PUT).  Of course, if we're looking
at sending back the entire Performance Timeline (which includes Navigation
Timing 2, Resource Timing, and User Timing), POST might be our only option
simply due to the large amount of data.

Mike  
Received on Monday, 9 September 2013 20:38:24 UTC

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