RE: ACTION-660: Input to BP2, on Personalization

Sean,
The mobile specific angle is that the cost of making the user enter
information is much higher in the mobile context.

I can verify that "X-" headers are used. Their use though is generally
proprietary, specific to service provider deployments, and not usually
something that the service providers would share except through their
internal developer/partner documentation (usually open to any who sign
up though). We do add value by clarifying that this is an option that is
useful, and that details are provided by the service providers. So if I
develop an application that becomes user-awareness-persistent through
use of a forwarded "X-" header, I have met the criteria of the
recommendation. W3C does not have to say in detail what those are;
developers learn it from being part of a community, and the details vary
between service providers anyway.

Other than these proprietary methods, other standardized methods are in
development e.g. in OMA and 3GPP, but not yet available in devices and
thus out of scope for BP2.

Re cookies, yes there is an assumption here (it will be more explicit in
the "how to do it") that cookies are a useful way to manage
personalization persistence. But the recommendation is written in
generic terms so that various methods of doing it will meet the
recommendation.

Re staying logged in, it's the same basic objective, to manage
statefulness. There again, various methods will work including cookies.

Best regards,
Bryan Sullivan | AT&T
-----Original Message-----
From: Sean Owen [mailto:srowen@google.com] 
Sent: Thursday, February 14, 2008 2:27 PM
To: Sullivan, Bryan
Cc: BPWG-Public
Subject: Re: ACTION-660: Input to BP2, on Personalization

On Thu, Feb 14, 2008 at 5:02 PM, Sullivan, Bryan <BS3131@att.com> wrote:
>  [bryan] These three recommendations address the basic ability to  
> minimize user effort in personalizing services. They should be easily

> testable, at least manually. The methods of implementing the  
> recommendations will be described. Those based upon standards will be

> specifically described. Those based upon standard extensions or even  
> proprietary methods (e.g. for the first, based upon "x-" headers as  
> typically used by network proxies) will be mentioned generally (as 
> types  of methods, but not with specifics). Either will suffice for 
> compliance  if they result in the general recommendation being met. We

> welcome  suggestions for other recommendations in this area.

[srowen] My $0.02 on the possible mobile-specific angles here if any
are:

- Yes, are there "X" headers that people really rely on in practice to
identify sessions and users? I think this is actually so. I know we do
some awful magic with MSISDN info we get as a surrogate cookie
sometimes. I *do* think we should be specific, or else we're not adding
much value here.

- The problem of not having cookies all the time to personally identify
a user (or are we kinda assuming cookie supports? oops, there's the ADC
ghost... but don't we need to have assumptions like this to get
anywhere?)

- The need to stay logged in and avoid re-entering credentials all the
time. Hmm, what can we say here?

Received on Thursday, 14 February 2008 23:09:58 UTC