Re: xpath string index function

On 4 Oct 2011, at 08:57, Andy Seaborne wrote:
> On 04/10/11 04:59, Axel Polleres wrote:
>> On 3 Oct 2011, at 23:39, Andy Seaborne wrote:
>>> I have prototyped:
>>> 
>>> STRBEFORE
>>> STRAFTER
>>> REPLACE
>>> 
>>> and didn't encountered anything unexpected.
>>> 
>>> I propose adding these as resolution of JB-7 (part 1)
>> 
>> 
>> +1 to add those

+1 from me too.

>> Note that in RIF we also had adopted these additional 3 ones:
>>  fn:string-join   ( possible name: STRJOIN )
> 
> CONCAT
> 
>>  fn:iri-to-uri    ( possible name: IRI_TO_URI )
> 
> +0 (AKA I don't mind but don't think it's important)

Agreed. I'm even slightly worried that it's a bad idea. SPARQL systems are supposed to work with IRIs

>>  fn:escape-html-uri ( possible name: ESCAPE_HTML_URI )
> 
> +0
> Again, I really don't see the transformation capabilities as very important but I can be persuaded otherwise.  All this "not IRIs" seems a bit historical.

Quite.

- Steve

>> Just wanted to ask whether we think those should also be added? (the advantage being that we'd be at least comaptible with the function set in RIF)
> 
> I'd like to understand the reasons they were added.  Was it for specific reasons / use cases , or based on an audit of F&O?  Compatibility is a reasonable reason.
> 
> 	Andy
> 
>> 
>> best
>> Axel
>> 
>>> 
>>>         Andy
>>> 
>>> On 29/09/11 12:11, Andy Seaborne wrote:
>>>> Proposal:
>>>> 
>>>> Add
>>>> 
>>>> SUBSTR_BEFORE (keyword for fn:substring-before)
>>>> SUBSTR_AFTER (keyword for fn:substring-after)
>>>> 
>>>> but keep using the XSD F&O operator to back the SPARQL keywords (i.e.
>>>> 1-based strings).
>>>> 
>>>> ---------
>>>> 
>>>> It seems to be this, or redo all the string operations as 0-based, and
>>>> have URI-calls to fn: string operations for the 1-based style.
>>>> 
>>>> -----
>>>> 
>>>> What about adding fn:replace as well?
>>>> REPLACE
>>>> http://www.w3.org/TR/xpath-functions/#func-replace
>>>> 
>>>> (I haven't checked the details of the replace language and rules to see
>>>> if its sufficiently close to all know programming languages to be easy
>>>> to implement via library calls - any one know?)
>>>> 
>>>> 
>>>> Andy
>>>> 
>>> 
>>> 
>> 
> 

Received on Tuesday, 4 October 2011 08:45:31 UTC