Re: Why not vCard?

Renato Iannella wrote:
>
> I am actually quite surprised that Portable Contacts is what the
> Contacts API WD is based on.
>
> All of the attributes listed in that Working Draft are pretty much what
> is in vCard (the industry standard with the widest deployment, from the
> IETF). I did not see the xtra social attributes listed (eg food, heroes,
> etc) used by PoCo.
>
> I tried to look thru the email archives to see why....and found a small
> discussion [1] with the argument:
>
> "As PoCo represents a larger set of Contact attributes, I'd like to
> propose that we make PoCo the basis of the DAP Contacts API"

Firstly I will say that nothing is set in stone here. There's a note in 
the spec specifically pointing to the fact that formats are currently 
under discussion. The spec is at the Working Draft stage specifically 
for this reason (and why we're not publishing it tomorrow).

It has hardly been a trivial discussion over the past few months and 
there are a _lot_ of discussions on the contacts format that we should 
use on the mailing list.

A number of web services use Portable Contacts and OpenSocial as the 
basis of Contact Information Sharing. Traditional Address Books, for the 
most part, use vCard. The Contacts API acts in the space between the 
device and the web and so choosing a device facing or web facing 
paradigm is going to result in someone losing data in transfer 
somewhere. Someone is going to be unhappy and so if we displease 
everybody equally but arrive at a workable solution, without 
re-inventing the formats, then perhaps we could live with that.

On top of that, the vCard standard does not provide a JSON mapping for 
its properties at present. It doesn't translate well to communication 
over the web at large. The DAP WG is defining WebIDL based APIs with the 
objective of being able to apply these definitions to REST-based API 
interactions in the future.

Finally, we have also specified the format based on real-world 
implementations around this concept, most notably the Mozilla Contacts 
project [3].

>
> Again, this "larger set" is not evident in the WD.

We're still deciding how to cover the larger set hence why we are still 
at the Working Draft stage.

My current plan was to stick with the basic Portable Contacts set 
included in the spec and then provide clear extension guidelines for 
developers to provide the larger set of OpenSocial properties as required.

>
> Was this the right decision to make? Given the provenance and industry
> acceptance of vCard?

Frankly we're between a rock and a hard place on this created by the 
incompatibility between Portable Contacts and vCard. The right decision 
is the wrong decision for someone else? Why is that? Why, in 2010, can 
we not agree on a single format to rule them all? Why am I losing my hair?

Portable Contacts aims to cover the fields of vCard (except for using 
e.g. archaic spellings) which provides a route to generating vCard data 
from an object provided by the API. On top of that the vCard spec also 
provides good extension guidelines. With the aid of a co-ordinated 
'Portable Contacts extensions for vCard' draft, or a company-driven 
implementation of the same thing, we might be able to make such a 
solution work and grow at its own pace.


>
> And, ironically, the Introduction [2] says "The API itself is agnostic
> of any underlying address book sources"

Right. Apologies if that is not clear. It means that an underlying 
address book can be in whatever format it likes but when it is served up 
via the Contacts API it assumes the format described in the spec. It 
means that data storage is not necessarily equal to the required 
input/output via the API.

- Rich

>
> Cheers
>
> Renato Iannella
> http://renato.iannella.it
>
> [1] http://lists.w3.org/Archives/Public/public-device-apis/2009Nov/0045.html
> [2] http://www.w3.org/TR/2010/WD-contacts-api-20100701/#introduction
>

[3] http://mozillalabs.com/contacts

Received on Thursday, 9 September 2010 08:46:09 UTC