[Bug 4841] [FS] Use of fn:subsequence in relation to normalization rules for filter expressions in FS

http://www.w3.org/Bugs/Public/show_bug.cgi?id=4841





------- Comment #10 from jmdyck@ibiblio.org  2007-10-30 04:31 -------
At meeting 337, I was directed to propose a solution that did not change the
results of static type analysis. This excludes all of the above suggestions
except #2 in Comment #5. Here is a more fleshed-out version of that solution.

(A)
At every occurrence of "fn:subsequence" in 4.2.1 and 4.3.2:
-- Change "fn:subsequence" to "fs:item-at".
-- Delete the third argument (always "1").

(B)
Create a new section:

  7.1.13 The fs:item-at function

    fs:item-at($sourceSeq as item()*, $loc as xs:double) as item()?

    The fs:item-at function returns the item at a specified position in
    a sequence.

  Dynamic Evaluation

    If $loc is numeric-equal to the position of some item in $sourceSeq,
    that item is returned. (This implies that $sourceSeq is non-empty,
    and $loc is numeric-equal to an integer between 1 and n inclusive,
    where n is the number of items in $sourceSeq.)

    Otherwise, the empty sequence is returned.

    The function is roughly equivalent to the following user-defined
    function.

        declare function fs:item-at(
            $sourceSeq as item()*,
            $loc as xs:double) as item()?
        {
            if ($loc div 1 eq 0) then
                fn:subsequence($sourceSeq,$loc,1)
            else
                ()
        };

  Static Type Analysis

    The static typing rules for invocations of fn:item-at depend on the
    syntactic form of the second argument.  By construction, this must
    be either $fs:last, $fs:position, or a NumericLiteral. Thus, those
    are the only cases considered by the following rules.

    [Then, from 7.2.13 / STA, move all but the last rule (and its
    preceding prose) to here, with the following changes:
        'fn:subsequence'       --> 'fs:item-at'
        (FN-URI,"subsequence") --> '(FS-URI,"item-at")'
        In each rule's conclusion, drop the function call's third arg
        (always "1").
    ]

(C)
In "7.2.13 The fn:subsequence function":

  Introduction:
    Drop it.

  Static Type Analysis:
    Everything except the last rule (the catch-all case) has been moved
    to 7.1.13.

    Change the last rule's preceding prose to: 
        The static type for the fn:subsequence is computed using prime
        and quantifier, which are defined in [8.4 Judgments for FLWOR
        and other expressions on sequences].

(D)
In "C.1 Functions and Operators used in the Formal Semantics", drop the
list item for fn:subsequence.

Received on Tuesday, 30 October 2007 04:31:25 UTC