Re: [xsl] About position() in xsl:for-each-group

[Colin Paul Adams <colin@xxxxxxxxxxxxxxxxxx>]

>>>>> "Christian" == Christian Rinderknecht <rinderkn@xxxxxxxxxxxx> writes:

    Christian> <xsl:for-each-group select="cities/city"
    Christian> group-by="@country"> <xsl:value-of select="position()/>
    Christian> </xsl:for-each-group>

    Christian> which prints the positions of each group as determined
    Christian> by the document order of the first item of each group,
    Christian> and

This returns the context position, which is defined as:

"the context position is the position of this item within the set of
initial items (that is, one item for each group in the population) in
population order, "

so effectively it is the position of the group within the sequence of
groups.


    Christian> <xsl:for-each-group select="cities/city"
    Christian> group-by="@country"> <xsl:value-of
    Christian> select="./position()/> </xsl:for-each-group>

    Christian> which prints the position of the first item in each
    Christian> group, i.e., 1.

The context item (.) is defined as "the context item is the initial item
of the group". This is a sequence of 1 item. So ./position() is giving
the context position within a sequence of 1 item, so it will always be 1.


    Christian> But

    Christian> <xsl:value-of select="@country"/>

    Christian> and

    Christian> <xsl:value-of select="./@country"/>

    Christian> return the same value.

Yes. Because you are asking for the country attribute:

a) of the first item in each group, and
b) of each item in a sequence of one (the first item in the group)

    Christian> Does this mean that, in general, an XPath expression
    Christian> "p" is NOT equivalent to "./p"? (Philip Wadler (2001),

The / operator changes the context item, so yes.

    Christian> in his paper entitled "A formal semantics of patterns
    Christian> in XSLT" at
    Christian> http://homepages.inf.ed.ac.uk/wadler/topics/xml.html#xsl-semantics
    Christian> wrote, about his formalisation of a fragment of XPath
    Christian> 1.0 that `It follows immediately that the patterns
    Christian> p/. and p and ./p are all equivalent.')

Expressions are not patterns - the latter are syntactically a subset
of the former, but the semantics of pattern matching are not those of
expression evaluation.

    Christian> Moreover, you say that "position()" (i.e., the context
    Christian> position) is not a property of a node, but the XPath
    Christian> 2.0 standard says that "It changes whenever the context
    Christian> item changes." The only related standard where I found
    Christian> the word "iteration" is in 14.1 of the XSLT 2.0
    Christian> Standard, about the notion of "Current Group". In
    Christian> particular, it is not found in XPath nor the Data
    Christian> Model, so I am not sure what you mean, since position()
    Christian> can be used as

    Christian> <xsl:value-of select="cities/city/position()"/>

    Christian> where I do not see any iteration (in a loopish sense).

That is correct. of course, the implementation may well involve a
loop, but it does not have to.

    Christian> Also, the constructor sequence in the
    Christian> xsl:for-each-group element could be evaluated in
    Christian> concurrent threads having a different context item,

Or concurrent processes. Or for that matter a sequential evaluation in
a different order from that of the sequence, results being
subsequently re-ordered to preserve the semantics.

    Christian> that is, there is no sequentiality implied when
    Christian> evaluating the sequence

Correct. That is, there is a sequential ordering of results that must
be observed, but no temporal evaluation order.

    Christian> and the iteration you refer to
    Christian> seems to me more a concept related to the serialisation
    Christian> of the output (the side effect of the evaluation). (The

Serialization does not necessarily take place.

-- 
Colin Adams
Preston Lancashire