Excerpts from Nina Jeliazkova's message of Tue Oct 12 13:44:07 +0200 2010:
> 
> Fine for me  - should we stick to literals or go for more complex constructs
> ?

I think that string literals should be suffficient. 

> BTW, in opentox.owl there should already be objects of type ot:Error, as
> defined in the errors proposal by NTUA.
> 
> Also, API - wise ,  how these should be accessed ?  I can think of two
> options:
> 
> 1) via /object/id/metadata   , then we need to invent metadata resources for
> objects other than datasets;

/object/{id} could return also ot:Info/Earning/Error (with exception of
dataset, where we can use also /dataset/{id}/metadata, if we do not want to
download the complete dataset).

> 2)  via specific links /object/{id}/info , /object/{id}/warning ,
> /object/{id}/error

Loks nicer from a REST point of view, but I am not sure if it is worth
to imlemement 3 additional methods for each service.

Best regards,
Christoph