> Ick!
> On the other hand, zillions for permutations of fields as separate schemas
> is also ick.

Exactly.  At least this way the client can (potentially) ask for any
arbitrary permutation without having to have a well known identifier for
it, or something even more hideous like dynamic schema URIs:

http://srw.o-r-g.org/schemas/dynamic/ead/eadheader/titlestmt/titleproper[1]

If it was a well known identifier, then that identifier might just as well
be a schema + xpath ... so either 'implementation' is possible.


> This is a big step along the road to mandatory profiling.  Few of us will
> implement it...

It could be implemented as many many schemas with the same results
currently, so I don't see that it's any further along the road to
mandatory profiling than we already are.


> Well, that might not be true.  Dropping the request into an XSLT transform
> might be easy.  I'm already running transforms to provide the requested
> schema anyway.

And sort uses XPath for specifying keys already too.  Once you have a DOM
structure for the record, doing things with it is pretty cheap.  (Of
course, /creating/ the DOM can be expensive!)

But if you have an XSLT engine, surely you must also have an XPath engine
somewhere in there to process the XSL ?



> So, if you'll give some thought to how this might be easily implemented
> (just exactly what the XSLT boilerplate would look like), I guess I might
> refrain from whining.

My implementation would be something along the lines of:

import xpath

...

try:
  recordPath = xpath.compile(request.recordXPath)
except:
  diag = SRWDiagnostic(XXX)
  diag.details = "Unparsable xpath in request"
  raise diag

rec = SRWRecord()

try:
  node = recordPath.evaluate(document)
  rec.recordXPath = request.recordXPath
  rec.recordData = str(node)
except:
  diag = SRWDiagnostic(XXX)
  diag.details = "XPath not valid for record"
  rec.recordData = str(diag)
  rec.recordSchema = "http://www.loc.gov/srw/diagnostic/1.0/"

response.records.append(rec)


Without an XPath implementation that's separate from your XSLT engine...

<xsl:stylesheet ... version="1.0">

  <!-- Wonderful library: http://xsltsl.sourceforge.net/ -->
  <xsl:include href="../markup.xsl"/>

  <xsl:template match="%%%REPLACE%%%" priority="2">
    <xsl:call-template name="markup:as-xml">
      <xsl:with-param name="nodes" select="."/>
    </xsl:call-template>
  </xsl:template>

  <xsl:template match="*" priority="1">
    <!-- Eat unwanted data -->
    <xsl:apply-templates/>
  </xsl:template>

</xsl:stylesheet>

(No guarantees the above will work, but I think it should...)



Rob

--
      ,'/:.          Rob Sanderson ([log in to unmask])
    ,'-/::::.        http://www.o-r-g.org/~azaroth/
  ,'--/::(@)::.      Special Collections and Archives, extension 3142
,'---/::::::::::.    Nebmedes:  telnet: nebmedes.o-r-g.org 7777
____/:::::::::::::.                WWW:  http://nebmedes.o-r-g.org:8000/
I L L U M I N A T I