Thanks Paul.

>>>
>>>Actually, I had time to play around with the SAX interface today. It turned out to be MUCH faster. I dropped the 1.5 hour conversion time down to 70 seconds! The number of calls through COM didn't turn out to be that big of a deal. I was really surprised by how well it worked.
>>
>>Paul,
>>
>>Does this mean that XMLTOCURSOR() is NOT a resource hog or a sllooowwww process?
>
>For small XML files, XMLTOCURSOR() seems pretty quick. I've been using it to process cursors passed via a SOAP call in an app - of course, the result set is pretty small (less than 100 records). For large XML files, it's unusably slow. I didn't really do any performance testing so I can't give you any specifics as to whether the performance is linear, or if it gets exponentially worse, or something else. I also didn't pay too much attention to the amount of resources it took; I'm sure it's pretty heavy since it has to load the entire XML document into memory. It does peg the CPU usage at 100% while it's running, though.
>
>I'm not using XMLTOCURSOR anymore to do the conversion. I'm using the SAX interface of MSXML. It raises an event everytime it hits a new XML tag, or information contained inside a set of tags. So, I just added a bit of code to check to see if the tag that it just processed was my "record" tag. If so, I append a new record. When it raises an event on one of my field tags, I save off the field name. Then, when it raises an event for a value I save this value into the new record I created, in the field name I saved.