RE: [security-services] AI-37 DSig Core change proposal

[Scott Cantor <cantor.2@osu.edu>]

> I think that this kind of freedom leads to madness. The 
> minimal restriction involved in having a single ID attribute 
> (that must meet XML rules) seems tiny compared to the 
> confusion that could result from having two different ID 
> attributes and having to explain the difference to every new 
> implementer.

Agreed, somewhat. I don't personally notice XML IDs very much. I sort of skim them when I'm reading because I know what they're for
in a spec.

> Can we just change the type of the existing AssertionID and 
> RequestID attributes to xsd:ID, or are there tools that will 
> break because they're not literally named "ID"?

No, the names don't matter. The reason we weren't planning on doing it now was that the value domain of xsd:ID is smaller than the
domain of xsd:string. That is, if we change it and narrow the value domain in 1.1, we break backward compatibility, because an older
AssertionID could fail to validate.

-- Scott