[sc34wg3] TMQL: Operators

[Lars Marius Garshol]

* Lars Heuer
>
> 4.11 Value Expressions
> ----------------------
> [33] infix-operator  ::= ...any in the predefined environment...
> [34] prefix-operator ::= ...any in the predefined environment...
>
> Well, while the explanation is short and catches the point (I love
> compact statements), the grammar should list at least every
> *predefined* operator, like <= => etc., since I assume these operators
> shouldn't be overloaded by the concrete implementation / environment.

I definitely agree that they should be here. In my opinion it's not  
acceptable to have an incomplete grammar that has to be filled in  
from other sources, even if those sources are within the spec itself.

I accept that Robert's argument about specifying this in one place  
only makes sense. However, one possible solution is to move the  
operators out of annex A and into 4.11 instead. There seems to be no  
compelling reason to have them in A, anyway.

Another issue is: what about operator precedence? Where/how is that  
defined? (tmql-operator-precedence) XPath encodes this directly into  
the grammar. We don't necessarily need to do the same, but we have to  
do it *somehow*.

> BTW: I hope, that the functions are defined somewhere more explicitly?
> Either in the TMQL document or in a separated doc. The current
> description (encoded in CTM) lacks of some information (i.e. how does
> the syntax for RegExes look like etc.).

 From Robert's reply I assume that there is general agreement that  
this must be done. I've listed this as an issue (tmql-define- 
functions) that is settled, but not implemented in the spec yet.

--Lars M.