Notation of values (like xs:decimal) is undefined?

[EskoDijk]

I didn't find this in the terminology section - define a reference to something?

FYI @marco-tiloca-sics (maybe we need to pass this remark to the shepherd?)

[marco-tiloca-sics]

Also CCing the Shepherd @jaimejim

[bsilverajan]

What if instead of "xs:decimal" or "xs:boolean", the document simply refers to the type as "decimal" and "boolean"? Would that improve readability and understanding?

[EskoDijk]

@bsilverajan Using the "xs:" notation (referring to XML schema) also provides the complete definitions of allowed values. So it would be unwise to remove that, since it could lead to interop issues. E.g. my client uses the "TRUE" or "t" boolean value while the server expects "true". (Or maybe server doesn't accept '0' or '1' while it should.)

So there just needs to be a formal reference.

[bsilverajan]

Thanks, @EskoDijk this was indeed the original intent to refer to xml schema definitions. I'll leave this issue open until I include a reference from the W3C XSD specs here to the draft.

Just a note regarding what you raised regarding interop issues: Providing the reference mitigates this somewhat but does not necessarily solve it completely as you noted: While the canonical representation accepts just {true, false}, the lexical representation can include {true, false, 1, 0}, according to the specs.

[EskoDijk]

@bsilverajan I think the XML schema specs may require parsing of any valid lexical representation; which would make sense for the use case here.

For example if the client adds &c.gt=23.000 this should be equally valid to &c.gt=23 or &c.gt=0023.0 right?

(If we would only accept canonical form, that's really important to point out!)

[bsilverajan]

That is exactly right, and I dont think the document should go into that level of detail at all and should just cite a reference to the W3C specification. Right now, the appendix provides some pseudocode to illustrate what the server could provide as a boolean.

[EskoDijk]

@bsilverajan I see in the Appendix examples of the xs:decimal that a client provides. Is there something for xs:boolean? (Provided by the client, or not?)

[bsilverajan]

@EskoDijk I mistyped sorry! I meant decimal, not boolean of course (as a response to what you mentioned regarding the validity of values that c.gt can accept for an example server pseudocode).

[bsilverajan]

@EskoDijk I'm drafting the following text in the terminology section as follows:

"xs:boolean" and "xs:decimal" types:
: Data types from XML Schema {{XMLSchema2}}, used in this document to describe boolean and scalar values in CoAP resources. The "xs:" prefix notation is used solely for type indication and does not imply the use of XML or XML Schema in protocol encoding.

where XMLSchema2 is a normative reference to https://www.w3.org/TR/xmlschema-2/

Does this look good to you?

[EskoDijk]

@bsilverajan Yes, looks good! Thanks.