Re: fn:foot, fn:truncate, fn:item-at

So, we already have two different existing ways for this, and now we are
adding a 3rd one !

I don't see any specification for  slice()  in the F&O document, maybe it
provides some useful functionality beyond what can be done with subsequence
and predicates.

For example, if:

slice($seq, -1, -$N)

  produces the reversed subsequence (starting from the last item, then this
may be useful.

Thanks,
Dimitre

On Wed, Dec 2, 2020 at 9:29 AM Michael Kay <mike@saxonica.com> wrote:

> The inspiration for slice() came from JSONPath, which itself appears to
> have got it from Python.
>
> We could devise custom syntax for the operation, for example $input[start
> ; end (; step)] but a function seems more powerful and just as usable.
>
> The main problem with relying on predicates for positional subscripting is
> the fact that the context changes in the predicate, which means that when
> you write things like
>
> (1 to $N) ! (first-names[.] || last-names[.])
>
> it doesn't do at all what you intended.
>
> Similarly $input[$N] doesn't do what you intended if $N is a node that
> needs to be atomized (classic pitfall in XSLT when you do
> <xsl:variable><xsl:value-of select="$N+1"/></xsl:variable> - though we
> won't train them out of that if we haven't already).
>
> In addition doing commonly-needed things like $input[position() mod 2 = 0]
> is just really clumsy.
>
> subsequence() doesn't have the problems of the context changing, but
> writing subsequence($input, $N, 1) to get the item at position N is
> terribly long-winded compared with the equivalent in other languages, and
> the fact that it allows non-integer values also complicates the
> implementation no end, as well as making errors harder to debug.
>
> Typical example today:
>
> <xsl:for-each select="first-names">
>   <xsl:variable name="p" select="position()"/>
>   <xsl:value-of select=". || last-names[$p]"/>
> </xsl:for-each>
>
> With slice:
> first-names ! (. || last-names => slice(position()))
>
> (OK, there's another solution using for-each-pair - but the more you use
> higher-order-functions, the more useful it is to deliver functionality
> using functions rather than custom syntax.)
>
> Michael Kay
> Saxonica
>
> On 2 Dec 2020, at 15:27, Dimitre Novatchev <dnovatchev@gmail.com> wrote:
>
> > slice($seq, -1)  to get the last item
>
> > slice($seq, $N) to get the Nth item
>
> > slice($seq, 1 to count($seq)-1) to get all but the last
>
> -10  !
>
> $seq[last()]
>
> $seq[$n]
>
> $seq[position() lt last()]
>
> All of these are shorter than the proposed expressions containing slice()
> ! Why bother learn a new name and why pollute the function-name space?
>
> This is just an example where we are going counter - Exupery and with
> close to no justification.
>
> We also have another good friend:
>   subsequence()
>
>
> Thanks,
> Dimitre
>
>
> On Wed, Dec 2, 2020 at 3:16 AM Michael Kay <mike@saxonica.com> wrote:
>
>> The latest version allows
>>
>> slice($seq, -1)  to get the last item
>>
>> slice($seq, $N) to get the Nth item
>>
>> slice($seq, 1 to count($seq)-1) to get all but the last
>>
>> and I thought that was probably good enough to make the individual
>> functions unnecessary.
>>
>> Would you agree?
>>
>> Michael Kay
>> Saxonica
>>
>> On 2 Dec 2020, at 10:30, Christian Grün <cg@basex.org> wrote:
>>
>> Hi Michael,
>>
>> In your first public XQFO proposal [1], I have found seen three
>> functions (fn:foot, fn:truncate, fn:item-at) that have not made it
>> into the latest specification documents. Are you still working on it,
>> or did you deliberately decide not to include them in the new draft?
>>
>> Thanks,
>> Christian
>>
>> [1] https://www.saxonica.com/qt4specs/FO/Overview-diff.html
>>
>>
>>
>
>
> On 2 Dec 2020, at 15:29, Dimitre Novatchev <dnovatchev@gmail.com> wrote:
>
> > slice($seq, -1)  to get the last item
>
> > slice($seq, $N) to get the Nth item
>
> > slice($seq, 1 to count($seq)-1) to get all but the last
>
> -10  !
>
> $seq[last()]
>
> $seq[$n]
>
> $seq[position() lt last()]
>
> All of these are shorter than the proposed expressions containing slice()
> ! Why bother learn a new name and why pollute the function-name space?
>
> This is just an example where we are going counter - Exupery and with
> close to no justification.
>
> We also have another good friend:
>   subsequence()
>
>
> Thanks,
> Dimitre
>
> On Wed, Dec 2, 2020 at 3:16 AM Michael Kay <mike@saxonica.com> wrote:
>
>> The latest version allows
>>
>> slice($seq, -1)  to get the last item
>>
>> slice($seq, $N) to get the Nth item
>>
>> slice($seq, 1 to count($seq)-1) to get all but the last
>>
>> and I thought that was probably good enough to make the individual
>> functions unnecessary.
>>
>> Would you agree?
>>
>> Michael Kay
>> Saxonica
>>
>> On 2 Dec 2020, at 10:30, Christian Grün <cg@basex.org> wrote:
>>
>> Hi Michael,
>>
>> In your first public XQFO proposal [1], I have found seen three
>> functions (fn:foot, fn:truncate, fn:item-at) that have not made it
>> into the latest specification documents. Are you still working on it,
>> or did you deliberately decide not to include them in the new draft?
>>
>> Thanks,
>> Christian
>>
>> [1] https://www.saxonica.com/qt4specs/FO/Overview-diff.html
>>
>>
>>
>
>
>