Re: calling for xproc pain points, requested features, etc

Geert Josten <geert.josten@dayon.nl> writes:
> (sorry, meant to sent this to the list)
>
> -----Oorspronkelijk bericht-----
> Van: Geert Josten [mailto:geert.josten@dayon.nl]
> Verzonden: dinsdag 10 januari 2012 8:54
> Aan: 'Norman Walsh'
> Onderwerp: RE: calling for xproc pain points, requested features, etc
>
> +1 for adding a select on p:for-each. It could work similar to xsl:variable
> and xsl:attribute. If there is a select, p:iteration-source is ignored (or
> better: disallowed). In other words: keep the syntax compact where possible
> and sensible..

That's an interesting idea. I'm not opposing it, but I think we also need to
keep an eye on how many "ad hoc" syntactic shortcuts we add to the language.
There's tension between keeping everything completely regular and consistent
vs. adopting every shortcut imaginable.

I have some concern that a story which says this is valid:

  <p:for-each select="//chapter">...</p:for-each>

but this is not

  <p:for-each select="//chapter">
    <p:iteration-source>
      <p:pipe step="somestep" port="result"/>
    </p:iteration-source>
  </p:for-each>

and that it has to be rewritten this way:

  <p:for-each>
    <p:iteration-source select="//chapter">
      <p:pipe step="somestep" port="result"/>
    </p:iteration-source>
  </p:for-each>

is confusing. OTOH, this clearly has to be forbidden:

  <p:for-each select="//appendix">
    <p:iteration-source select="//chapter">
      <p:pipe step="somestep" port="result"/>
    </p:iteration-source>
  </p:for-each>

Just a few thoughts...

                                        Be seeing you,
                                          norm

-- 
Norman Walsh
Lead Engineer
MarkLogic Corporation
Phone: +1 413 624 6676
www.marklogic.com