Re: The xs:anyURI problem…

Wendell Piez <wapiez@wendellpiez.com> writes:
> I wonder how many users actually need or want the old ('wrong') behavior.

I wonder how many users are unaware that they’re relying on it!

> Whatever we do about namespace-rename, this makes me lean towards a strict approach of effectively deprecating and discouraging this.

I’m not sure what you want to deprecate or discourage. The namespace-rename step just needs to be changed so that the $from and $to options are xs:string instead of xs:anyURI, I think.

> Error handling when URIs fail might also be enhanced/enhanceable with diagnostic info such as relative-part and base URI.

Do you have an example where you think it’s currently insufficient?
The message today is along the lines of “Invalid URI: %gg” which is about as clear as I can imagine making it.

> I'd also like to know if you, Achim, George and others feel (as I do) that the stricter model is more robust and secure. (And as you say, one can always pass a string or heck, even a function?)

I’m not sure about robust and secure, what I see is that it’s more consistent. I think a user calling

  x:myStep

should not have to know or care if x:myStep is a user defined step or a built in atomic step. They shouldn’t be able to tell.

> Thanks and happy solstice, today is actually longer, in the 'more daylight' sense --

All the days in 2025 have been long, my friend. Here’s to 2026 being bettter!

                                        Be seeing you,
                                          norm

--
Norm Tovey-Walsh <ndw@nwalsh.com>
https://norm.tovey-walsh.com/

> What if this weren't a hypothetical question?

Received on Wednesday, 24 December 2025 12:44:50 UTC