tysontune

Perfect!  I blanked entirely on output buffer.  Thanks!

I'm building an application that is essentially a back end supplying xml to both flash and PHP.  The PHP front end is created through XSLT.  The xml files are output dynamically by setting the text/xml header in PHP.  Since the PHP page that does the xsl transform is on the same server as the xml producing back end, simple reading the xml files gives me the actual code contents of the file, not the compiled xml output.  To get around this, I have created a CURL function set that grabs the xml through http, which get's the correct data.  It also does this for the xsl files which are created with php files and output as text/xml as well. I'm married to the xml return for this one since that is how the data gets into the Flash component, and the database caching system I have setup returns the xml as well. So here's the question.  Am I creating too many requests to the server for this solution to scale?  In essence, you go to a page which calls two other pages via http, then does the transform.  This seems counter intuitive, but it is in widespread use with httprequest and ajax all over the web. The alternate is to create the output each an xslt is called, but this seems like more strain on the server than simple requests. Any idea what will scale better?