How bout the XML DOM Parser?

>I'm parsing some XML from an outside source using string concatenation to populate temporary tables for reporting. It works pretty well...
>
>... until you throw it XML in the tens of thousands of lines! I ran the code profiler against a sample report and found most of the trouble was from STREXTRACT() lines of code which were run many thousands of times. The whole process takes 45 seconds, and one of the STREXTRACT() lines uses 19 seconds to run about 10,000 times.
>
>I'm thinking of a few ways to speed things up - I was going to try using ALINES to throw the XML into an array and process it row by row, reducing the length of the string to be manipulated. Unfortunately that would mean I'd be limited to files of 25,000 lines (due to array limitations). I'm also concerned that multi-line fields would create problems.
>
>I've read that the XMLtoCURSOR stuff takes longer to run than direct string manipulation, so that's probably not a good direction to go in (in addition, this XML is sans DTD).
>
>Any other ideas?
>
>TIA