Re: [DM30] namespace names defined appears to be non-normative

Firstly, there is no WG (this is therefore a personal response). While the status section of the document invites readers to report errors, the document is now a final Recommendation and the time for suggesting improvements has long passed.

Secondly, the section you cite is carefully worded (and was discussed at length) to give implementors advice about how to deal with the conflicting requirements of other published specifications, concerning the exact definition of what is allowed in a namespace name. The solution is imperfect, but it can't be resolved without changing several other established W3C and IETF specs, and the final advice is unambiguous: 

implementations may reject character strings that are not valid URIs or IRIs, but they are not required to do so

If there were still a WG and if it looked at the section again I don't think it would want to change that advice.

My personal advice is: if you're a user, stick to valid URIs. If you're an implementor, be liberal in what you accept.

The reason we allow implementations to reject character strings that are not valid URIs or IRIs is that we want to allow implementations to use third-party libraries (for example, XML parsers or XSD 1.0 schema processors) that impose such restrictions.

Michael Kay
Saxonica

> On 27 Jun 2018, at 07:04, Mukul Gandhi <gandhi.mukul@gmail.com> wrote:
> 
> Hi all,
>    In the section, "2.6 Namespace Names" of the spec "XQuery and XPath Data Model 3.0", the language used in most of the section 2.6 appears to be non-normative (except for very few phrases) and is not very much helpful for implementation & inter-operability.
> 
> Is it possible, that WG re looks at the section I've pointed in this mail, for an improved description?
> 
> 
> 
> -- 
> Regards,
> Mukul Gandhi

Received on Wednesday, 27 June 2018 08:01:42 UTC