Datalink vocabulary

[Norman Gray]

Markus, hello.

On 2014 Sep 24, at 08:07, Markus Demleitner <msdemlei at ari.uni-heidelberg.de> wrote:

> On Tue, Sep 23, 2014 at 09:29:55AM -0700, Patrick Dowler wrote:
>> General question about this field: how should a client interpret a
>> null value for the semantics field? Normally this means "unknown" but
> 
> My take: The interpretation is "the data provider messed up".  That's
> because it's not unlikely that that's what happened (and for that
> reason I'm against overloading NULL as self; it's just too easy to
> get a NULL by accident and self is too important a category).

I agree with that.

> I'm not sure whether we should formally require non-NULL here; on the
> one hand, requiring it would let a validator (*obligatory bambi eyes
> all around*) flag a hard error if it is.  On the other hand, data
> providers will mess it up anyway, and clients should somehow
> gracefully fail on NULLs here (I'd say we should recommend just
> swallowing rows with NULL semantics; that'd let us use this as an
> extension mechanism, too, if need be).

Forbidding NULL is the more conservative decision, and to my eyes perfectly reasonable.  It can always be loosened up in future if there's a real need for NULL.

All the best,

Norman


-- 
Norman Gray  :  http://nxg.me.uk
SUPA School of Physics and Astronomy, University of Glasgow, UK