Re: ACTION-2211: Redefine terms via xdm instead of xpath[1, 2]

On a related point:

>  * for a calculate expression, the result is the empty sequence;

Isn't this too XPath 2 oriented? Wouldn't it be better to set the node  
value to the empty string?

Steven

On Wed, 06 Feb 2019 11:34:09 +0100, Steven Pemberton  
<steven.pemberton@cwi.nl> wrote:

> OK, how about:
>
> "Types in XForms are principally used to check the validity of strings  
> provided as values to nodes.
> However, these types can also be used by the expression language when  
> evaluating expressions using nodes; the XForms processor should  
> therefore provide the typed values of >nodes to the expression  
> evaluator, and not just the strings.
>
> If the evaluation of an expression fails because of a type error:
> * for a calculate expression, the result is the empty sequence;
> * for a constraint expression, the result is false();
> * for other bind computed expressions, the result is not applied to the  
> node;
> * for all other expressions, an xforms-expression-error event is  
> dispatched to the element containing the expression."
>
> Steven
>
> On Wed, 06 Feb 2019 07:32:23 +0100, Erik Bruchez <ebruchez@orbeon.com>  
> wrote:
>
>> This is my response to "ACTION-2216: XForms: Erik to review 2nd part  
>> and see if Steven's understnding is right, if there is too much text in  
>> the spec".
>>
>>> The other is section 6.2 Typed Values
>>> https://www.w3.org/community/xformsusers/wiki/XForms_2.0#Typed_Values
>>> which I now realise I don't understand the purpose of entirely.
>>>
>>> I think the argument is
>>>
>>> "Types in XForms are principally used to check the validity of strings  
>>> provided as values to nodes. However, the expression language may also  
>>> use those types during the evaluation of expressions using those nodes.
>>>
>>> If the evaluation of an expression fails because of a type error:
>>> * for a calculate expression, the result is the empty sequence
>>> * for a constraint expression, the result is false()
>>> * for other bind computed expressions, the result is not applied to  
>>> the node
>>> * for all other expressions, an xforms-expression-error event is  
>>> dispatched to the element containing the expression."
>>>
>>> Do you think there is any other part of that section that is normative  
>>> that I have missed?
>>
>> I don' think that's quite enough. We should also tell how "the  
>> expression language may also use those types during the evaluation of  
>> expressions using those nodes" in case >>there is not a type error.  
>> Mainly, this means that with XPath 2, the processor must expose the  
>> typed values of nodes when nodes are atomized or accessed with the  
>> `data()` >>function.
>>
>> -Erik