Re: label-elements proposal

On 9/6/07, Alex Milowski <> wrote:
> Based on what I've heard so far, here is the most inclusive proposal:
>  * We add an option called "scheme" that defaults to "integer".  All other
>    values are implementation defined.  A minimally conforming implementation
>    must support "integer".

please call it "decimal"

>    If someone wants a hexi-decimal number, then can:
>       * implement a scheme called "hexi-decimal"
>       * use XSLT to transform the decimal integer to hexi-decimal.
> * For the "integer" scheme, numbering starts at 1 and increases by one
>    every element regardless of whether the attribute is set on the element.
> * There is no uniqueness constraint.
> * We add an option called "replace" that takes a boolean value.  A true
>    value means the attribute value is replaced if the attribute already
>    exists on the element.  The default value is "false".
> I actually really like Murray's suggestion of the "scheme" option because that
> lets me add a "UUID" scheme and generate unique labels across documents.  I'd
> be quite happy if we added a "UUID" scheme as an optional scheme in our
> specification.

I prefer not to go there
I prefer to keep the simple as short as possible
'scheme' can have the default value of 'decimal'
a conformant implementation must handle scheme 'decimal'
all other value MUST be in a qname-but-not-ncname and is implementation defined


Innovimax SARL
Consulting, Training & XML Development
9, impasse des Orteaux
75020 Paris
Tel : +33 9 52 475787
Fax : +33 1 4356 1746
RCS Paris 488.018.631
SARL au capital de 10.000 

Received on Thursday, 6 September 2007 20:46:50 UTC