Re: Agenda: Web Annotation Teleconference 9 Sept 2015

Hi, folks–

On 9/8/15 9:18 AM, Benjamin Young wrote:
> On Mon, Sep 7, 2015 at 3:22 AM, Ivan Herman wrote:
>>     On 04 Sep 2015, at 18:19 , Robert Sanderson wrote:
>>      * Revert the @id and @type change, or at most have it in a third,
>>     optional context
>
>     I am not sure there is, indeed, a problem, see my separate mail on
>     the subject:
>
>     https://lists.w3.org/Archives/Public/public-annotation/2015Sep/0082.html
>
>     at this moment, unless there really *is* a major problem, I do not
>     see any new issue/evidence that would warrant us to revert the
>     decision. I certainly do not want to make this decision without
>     further discussions.
>
>     So -1 again (at least for now)
>
>
> I'm +1 to reverting this change, actually.
>
> In the case of Hypothesis' JSON we (currently) use `id` to be the
> non-URL `id` of within our system. Much of the rest of our JSON is
> already informed by the spec and "magically" upgrades if I add the
> context to it.
>
> See http://hypothes.is/api/search for a list of examples you can use in
> the JSON-LD playground. :)
>
> If I add the current context file that maps `id` to `@id`, then the `id`
> field would throw an error as it's not a URL, but if it remained `@id`
> then it's a "new thing" (to our system at least) and we'd populate it
> accordingly--vs. having to change `id` and the code that currently
> depends on it.

Couldn't we just use a different attribute name altogether? It's not 
intuitive to me that `id`/`@id` is typed as a URL.

In HTML/SVG/etc., for example, the `id` attribute could contain a URL 
(so long as it didn't have a space), but it would be pretty unusual.


> Likely other people are in the same boat, or they've used `id` and
> `type` within their system in non-JSON-LD related ways (I certainly have
> a stack of apps that use both of those in very idiosyncratic ways).

This suggests to me that maybe we shouldn't use common and overloaded 
names like `id`, `type`, and `role`.

(As a general rule of thumb, names that are used as common attribute 
names in HTML, but which have different semantics or syntax, might be 
best avoided; there's not much overlap, so it should be easy.)


> Maybe other devs out there have other examples? Would be nice to have
> some scenarios to discuss.

Indeed.


Regards–
–Doug

Received on Tuesday, 8 September 2015 16:26:36 UTC