CODEWITHSUNDEEP
xsltxpath
Tim
Tim

Reputation: 1023

best practice for efficient xpath selectors

Are there some general tips or best practices on how to create more efficient xpath selectors in xslt?

For example, say you have a DocBook XML document that only contains first-level sections, like this:

<chapter>
  <section><info><title>my first section</title></info> ... </section>
  <section userlevel="expert"><info><title>my second section</title></info> ... </section>
  <section><info><title>my third section</title></info> ... </section>
</chapter>

I think that in nearly all cases a selector like //d:section(1) is worse that one like /d:chapter/d:section(2). Can I generalize that and say that the more selective the selector the faster it will be?

For another example: If you want the second section element, you could pull it from a sub-tree of all sections or select it directly with //d:section[@userlevel]. This selector uses the attribute which is more selective but would it be faster?

Rules of thumb of what to do (or to avoid) when using XPath?

Upvotes: 2

Views: 1299

Answers (1)

Martin Honnen
Martin Honnen

Reputation: 167716

I think you need to look at a particular XSLT or XPath or XQuery processor or even at a particular processor and a particular tree model and then run tests. For instance //foo so so common that it might be optimized in an implementation.

When I run Saxon 9.5 Java HE with the XQuery //section ten times on your input sample I get "Average execution time: 1.007739ms" while with /chapter/section I get "Average execution time: 1.443799ms".

Upvotes: 2

Related Questions