W3C home > Mailing lists > Public > public-annotation@w3.org > August 2015

Re: [web-annotation] Should we systematically use typing in the model (and in the examples)?

From: Rob Sanderson via GitHub <sysbot+gh@w3.org>
Date: Thu, 06 Aug 2015 17:21:33 +0000
To: public-annotation@w3.org
Message-ID: <issue_comment.created-128449552-1438881692-sysbot+gh@w3.org>
>From an RDF point of view, no. From a developer point of view, I 
(personally) would prefer to be told what sort of thing the object is,
 rather than having to guess by introspecting on the object's 
properties.  Especially if properties or patterns beyond those of OA 
are used.

For example:
```
{ "value": "paris" }  or in Turtle  [ rdf:value "paris" ]
```

EmbeddedContent? FragmentSelector? HttpRequestState? I can tell from 
the relationship and structure, but it's more work than just being 
told :)


Worse:
```
{ "start": 100, "end": 400 }
```
Could be legitimately either a TextPositionSelector or a 
DataPositionSelector, and the client would do different things for 
those two classes.
In this case I need to be told which.
```
[ "members": [1,2,3] ]
```
A oa:Choice or a oa:List?  Again, very different behavior.

Programming defensively, one would check for the properties that are 
absolutely required and not rely on type ... unless you have to. But 
it's much easier (again IMO) to do a @type value to internal class map
 when creating the structure in your environment.


-- 
GitHub Notif of comment by azaroth42
See 
https://github.com/w3c/web-annotation/issues/61#issuecomment-128449552
Received on Thursday, 6 August 2015 17:21:36 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 18:54:39 UTC