Suspected error in csvw.jsonld for groupChar definition

[RickMoynihan]

In the published JSONLD vocabulary definition here:

https://www.w3.org/ns/csvw.jsonld

We see the following definition for groupChar:

https://github.com/w3c/csvw/blob/195b01a59c49b196bf9a233d7292e029a7c720f5/ns/csvw.jsonld#L167-L170

I'm not 100% sure but I believe that "NumericFormat,xsd:string" is a mistake, and that it should instead be expressed as an array like this: ["NumericFormat", "xsd:string"].

[gkellogg]

From Tabular Data Model:

groupChar A string whose value is used to group digits within the number. The default value is null. If the supplied value is not a string, implementations must issue a warning and proceed as if the property had not been specified.

Examples are '.', and ','.

NumericFormat is defined only in the vocabulary definition is in the domain of groupChar, not in the range.

Additionally, as pointed out, the "NumericFormat,xsd:string" , while a valid string, is clearly inappropriate as an @type. The proposed solution is to update the namespace document to have the value be simply xsd:string.

[RickMoynihan]

Ah yes that makes much more sense! I didn't look too much at what the intention was supposed to be; it was a more immediate practical problem where the CSVW vocabulary definition actually failed to parse as JSON-LD in titanium.

Thanks again 🙇

[barthanssens]

Hmz, it looks like the published context is still (or again ?) the older version: https://www.w3.org/ns/csvw.jsonld has the incorrect groupChar type ...

[gkellogg]

@iherman this was merged from #890 into gh-pages, but never seems to have been pushed to /ns. Can you take care of this?

[iherman]

Done.