XSLT performance issue

[JontyMC]

Is it better to specify a node directly?

i.e. is:

/firstlevel/secondlevel/somenode

better (or quicker) than saying:

//somenode

Thanks,

Jon

 

[3143]

I think it'd would be better, if you just writw the node as //somenode. That allows some mistakes and changes and it goes straight the address you wanted.

 

[JontyMC]

I am thinking more in terms of performance and good practice though. Does the XSLT processor find a node more quickly with the first method or is it indistinguishable?

 

[jel]

Consider this simplefied xml-document:

<library>
 <books>
  <book>
   <writer>Lowry</writer>
  </book>
 </books>
 <cds>
  <cd>
   <musician>Leadbelly</musician>
  </cd>
 </cds>
</library>

The x-path expression "//musician" forces the parser to check if a musician-node exitst in:
- library
- books
- (every) book
- writer
- cds
- (every) cd
The x-path expression "library/cds/cd/musician" forces the parser to check if a musician-node exitst in:
- (every) library/cds/cd
You bet there is a difference in performance, depending on the number of books you have.

As for good practice: suppose the xml is a bit more complicated:

<library>
 <books>
  <books_about_musicians>
   <book>
    <writer>Lowry</writer>
    <musician>Robert Johnson</musician>
   </book>
  </books_about_musicians>
 </books>
 <cds>
  <cd>
   <musician>Leadbelly</musician>
  </cd>
 </cds>
</library>

The x-path expression "//musician" might get you some unexpected results.

Conclusion: be as specific as you can.

 

[chiph]

XPath isn't terribly efficient anyway -- it (usually) does a straight linear search, so if your document gets over 1000 "records" you'll see search time skyrocket.

Chip H.


____________________________________________________________________
If you want to get the best response to a question, please read FAQ222-2244 first