Re: [Qexo-general] Speed of execution

[Per Bothner]

Eric Safern wrote:
> Per's suggestion is to pre-calculate the counts for the different types,

No, my suggestion is for each fortune type, create (in advance)
a separate cookie-file:
/var/www/html/fortunes-1.xml
/var/www/html/fortunes-2.xml
...
/var/www/html/fortunes-n.xml
and then do:

$cookies := document(concat("/var/www/html/fortunes-1",$tp,".xml"),
       $min := 1,
       $max := count($cookies)
return  $cookies [ Rand($min,$max) ]

> Note this test should run *faster* than the final code - I'm not loading 
> the count data, or
> generating the random number.

But you're still reading the entire /var/www/html/fortunes.xml,
and creating a big data structure for it.

Of course we want Qexo to do this quickly - and quite likely
there are things we could relatively easily improve.  Still,
if you have large data sets and need to improve performance,
you need to avoid having to process the entire data set,
typically by using index files, or partition the data (as I
suggested here).

> I'm not sure I understood your final suggestion.  If I pre-read the 
> entire document in the server
> init code, can I tell XQuery to run the query against the document *in 
> memory*?  How
> would that work?

For example something like this:

class Cookies
{
  static final TreeList cookies = Document.parse("...");

  public static TreeList getCookies () { return cookies; }
}

Then in your Qexo program replace:
  document("/var/www/html/fortunes.xml")
by:
  Cookies:getCookies()

--
        --Per Bothner
address@hidden   http://www.bothner.com/per/