Re: [xsl] position last and attributes

Subject: Re: [xsl] position last and attributes
From: Ihe Onwuka <ihe.onwuka@xxxxxxxxxxxxxx>
Date: Thu, 20 Sep 2012 13:36:34 +0100
On Thu, Sep 20, 2012 at 1:24 PM, Wolfgang Laun <wolfgang.laun@xxxxxxxxx> wrote:
> On 20/09/2012, Ihe Onwuka <ihe.onwuka@xxxxxxxxxxxxxx> wrote:
>> On Thu, Sep 20, 2012 at 12:30 PM, Andrew Welch <andrew.j.welch@xxxxxxxxx>
>> wrote:
>>>> <A 1st="1" second="2" third="3" fourth="4" fifth="5"/>
>>>>
>>>> now tell me what are the first and last attributes of  A.
>>>
>>> Ok, but first (as hopefully this will explain it):
>>>
>>> 1. What is the position() of @third ?
>>>
>>> 2. Why do:
>>>
>>> @second/position()
>>>
>>> @third/position()
>>>
>>> both return 1?
>>>
>>
>> No. I'm moving on from the orthogonal conversation to one that is
>> conducted on my terms since you are the one trying to understand me.
>>
>>  <A 1st="1" second="2" third="3" fourth="4" fifth="5"/>
>>
>> 1.  what are the first and last attributes of  A.
>> 2. what are A/@*[1] and A/@*[last()]
>>
>> Is there a definitive  answer to 1 (I say no).
>
> Correct, because the question does not make sense: you are asking
> about non-existing properties of a *set*.
>

So if we ask the same question of it's representation and get an
answer what do we make of that answer (rhetorical).

>> Are there definitive answers to 2 (Well a processor will always give
>> you one if the set is not empty).
>
> No - we know that this is implementation dependent.
>
>>
>> Is A/@* a representation of the attributes of A. (I say yes).
>
> It has the same cardinality. All values of this representation are
> contained in the other representation. There is a bijection between these two.
>
>> Is A/@* a faithful representation of the attributes of A?
>
> No: it has lost the set-ness.
>
>>
>> If A/@* is a faithful representation of the attributes of A why does
>> it yields answers to questions that the original representation can't
>> answer.
>
> Because the transformation has added new properties.
>

So it's not a faithful representation.

>>
>> So what if A/@* is not a faithful representation. Should I present the
>> answers it gives me as a universal truth.
>
> No, it's the truth about the transformation process from one to the other.
>

so you may get paradoxical answers from such a process, which was the
point of the  post that started this thread. Is such an observation
useful. I think so.

PS people that care about ordering still use relational databases.

Current Thread