Re: Definition of user agent?

Hi John,

You wrote:

> Colleagues,
>
> What is the significance of the addition of this sentence to the
> definition of user agent:
>
>> This standard applies to user agents that (1) can access the general
>> browsable Web; (2) have a user interface that satisfies the
>> requirements in Determining User Preference in the [TRACKING-DNT]
>> specification; (3) and can implement all of the [TRACKING-DNT]
>> specification, including the mechanisms for communicating a tracking
>> status, and the user-granted exception mechanism.

This was discussed at the Sunnyvale meeting, and (IIRC) came out of the
browser breakout session and a subsequent breakout with browser,
industry, and privacy folk.

In those sessions we tried to come up with wording to scope the DNT work
to user agents that consumers typically use to browse the web. The goal
was to have a narrow enough definition to exclude HTTP-using networked
appliances (a refrigerator, say, or an HTCPCP-compliant coffee pot),
while being broad enough to include not only today's browsers and apps
which use embedded browser engines (such as UIWebView on iOS) but also
as-yet-unimagined applications future users will typically use to browse
the web of tomorrow.

> Does this mean a user agent is not compliant if it doesn't implement
> the UGE mechanism?

No, *can implement* is not the same as *does implement*. IIRC this "can"
wording was intended to exclude network intermediaries which do not
embed a JavaScript engine and lack user-facing UI (such as routers that
automatically add DNT:1 to all outgoing traffic).


Ted

P.S. I didn't participate in putting the DAA document together; the
above comments are simply based on the various discussions at the
Sunnyvale F2F.

Received on Monday, 8 July 2013 17:28:33 UTC