On Monday, October 15, 2012, Michael Kay <mike@saxonica.com> wrote: > >> As much as I agree with you about the evil of QNames in content, I don't really think it was the job of the Namespaces in XML spec to prevent this.
> > It was absolutely the job of the Namespaces in XML spec to get this right, and it failed dismally. Firstly, it got the layering wrong: naming is fundamental and can't be layered on top of the core as if it were an application. Secondly, it refused to describe an object model, and thus failed to make it clear whether prefixes were information-bearing or not (rather as XML itself failed to make it clear whether CDATA boundaries are information-bearing).
OK I wasn't really making a point at that fundamental a level. Amelia's points seemed geared towards a presumption of scoped prefixes. I agree that if the prefixes were instead invisible to the app layer, numerous problems go away. Doesn't really matter anyway for MicroXML interest because prefixes are entirely gone daddy gone.
