Re: [xsl] Kosher XSLT 3.0 numbering solution?

["Sewell, David R. (drs2n) dsewell@xxxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

A two-pass solution was one of the things that occurred to me. But "cheating" 
with the global variable update makes the single pass easy enough that I didn't 
bother. Certainly for more complex situations it would be preferable.

David

On Thu, 1 Sep 2016, Graydon graydon@xxxxxxxxx wrote:

> On Thu, Sep 01, 2016 at 03:48:45PM -0000, Michael Kay mike@xxxxxxxxxxxx
> scripsit:
> > > On 1 Sep 2016, at 16:27, Liam R. E. Quin liam@xxxxxx
> > > <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx> wrote: But surely it's not
> > > that hard to write a recursive template or function that finds the
> > > previous page break and adds one to it where needed?
> >
> > Actually it's quite tricky if you want to have a function/template
> > that both returns a modified version of the source tree with new
> > numbers inserted, and also returns the latest number used. Returning
> > two results using maps is possible, but not easy.
> >
> > Computing the numbers is ideally done by a fold operation that
> > processes all the nodes in the tree in document order, and XSLT 3.0
> > accumulators are essentially syntactic sugar for such a fold
> > operation.
>
> With XSLT 2.0, every time I had to worry about this sort of thing --
> table spans, numbering legislation, complex figure numbers -- I found it
> was much better to do an initial pass or passes to calculate values and
> to attach those values to the elements of an instance of the source
> document as attributes.  A subsequent pass did the manipulation on the
> modified source document instance using the pre-calculated values.
>
> XSLT 2.0 and subsequent makes very this easy; you can stick each pass in
> a variable, and apply-templates selecting that variable for the next
> pass.
>
> It might not meet your sense of elegance, but I've found it a good way
> to keep the complexities distinct.
>
> -- Graydon

--
David Sewell
Manager of Digital Initiatives
The University of Virginia Press
Email: dsewell@xxxxxxxxxxxx   Tel: +1 434 924 9973
Web: http://www.upress.virginia.edu/rotunda