[foaf-dev] [foaf-protocols] strawpoll: migrate from foaf:holdsAccount to foaf:account?

[Melvin Carvalho]

On Thu, Aug 27, 2009 at 1:29 PM, Simon
Reinhardt<simon.reinhardt at koeln.de> wrote:
> Dan Brickley wrote:
>> I lean towards saying we'll migrate from foaf:holdsAccount to
>> foaf:account. With the rise of all these Web 2 sites it is increasingly
>> important to be able to relate a description of a person to descriptions
>> of their numerous online accounts; the holdsAccount property name is
>> just that little bit too awkward there for comfort.
>
> Uldis Bojars and I talked about this last night and I feel I have to object to this change. I don't see a problem with the current property name (Uldis actually likes it more because it is more easily understandable) and changing it seems to lead to more problems than benefits. We know that overall the naming style in FOAF isn't very consistent and there are many other properties that could be renamed, more urgently than foaf:holdsAccount. But when changing a term name in a vocabulary as widely deployed as FOAF you have to consider the consequences very carefully.
>
> First of all, how many of the FOAF files out there use foaf:holdsAccount? Which of those can be changed easily by contacting people?
> While I'm in favour of renaming SIOC's sioc:User to sioc:UserAccount (because that one is really misleading and has been misunderstood in the past) the landscape of SIOC deployment is very different. SIOC is not yet as widely deployed and most of it is either centralised in large sites or export plugins for various web software, so changes in published data will generally be doable. For FOAF though you have tons of hand-crafted FOAF files out there (which I expect to make use of this property a lot) and it has been in existance for so long that it's hard to track down all the software making use of it.
>
> But that's only the publisher side. And since, as Dan said, foaf:holdsAccount wouldn't get phased out but foaf:account just introduced as a new term, old published data would still remain valid. But what about the consumer side? What apps are out there that expect the property foaf:holdsAccount and just wouldn't react to foaf:account? Do they use reasoning or not? Uldis also pointed out that there might be software that was part of research projects which have been concluded and so changes to that software are unlikely. The question is then, how relevant is such software?
>
> The last thing we talked about was term stability. foaf:holdsAccount is stated to be an unstable term. But what does that mean? That you're unsure it's the right way of modelling things? That you want to see how people adopt the term? I think such a statement can be a deployment observation at best. The term "testing" would then mean you throw this term into the ring and see if it works well in the wild. "unstable" would mean you think it's the right way of modelling things but you want to see if people actually make use of the term and if they use it properly or in an unexpected way or if they don't like the way it is defined. But I don't think you can use this as a basis for deciding if you can easily remove / change the term or not. You have to look at the actual deployment for only that will tell you what the term truly means. And if it is widely deployed and accepted, then the term isn't unstable anymore de facto.

Surely unstable means subject to change?  Anyone implementing a
version 0.9 of the spec or less, really ought to suspect there *might*
be some changes before all is said and done.

I personally look forward to the day when foaf-project.org broadcasts
machine readable spec changes using SPARUL, my reasoning and
reputation engine decides to auto approve the changes, and propagates
them through my system, as well as relay the changes to downstream
nodes.

>
> Dan, you also said that translating multi-word properties can be tricky. Well, the label doesn't have to be bound to the term name and I'd say, just be creative with the translations. :-)
>
> Regards,
>  Simon
> _______________________________________________
> foaf-protocols mailing list
> foaf-protocols at lists.foaf-project.org
> http://lists.foaf-project.org/mailman/listinfo/foaf-protocols
>