Is the parsing a significant portion of the time spent? Is there any kind of profiling tool for Flex?
I figured it this way.any XML format is going to spend roughly 50% more time on the wire than a binary format for larger chunks of data. If you're passing one field, say "<varname>value</varname>" it's not going to matter very much because packet overhead is going to be more significant than the extra 30 bytes of XML junk. If you're passing a record with 4,000 bytes of data, none of it in JPG's, then I am guessing the binary formats are going to spend less time on the wire. I am assuming (without measuring this is dangerous) that most people writing these applications are doing it like I am, for end-users at home on a DSL connection with 500-4,000 kilobits per second download speeds. With a large user population, the server bandwidth might also make binary objects economical. I didn't think a modern person computer would spend much time parsing the data, like 1,000th the time it takes to send the data over the wire. But, I did not measure it and the one rule of optimizing for speed that has served me well over the last two decades is always measure before you optimize. _____ From: flexcoders@yahoogroups.com [mailto:[EMAIL PROTECTED] On Behalf Of wayneposner Sent: Tuesday, April 24, 2007 9:30 AM To: flexcoders@yahoogroups.com Subject: [flexcoders] Performance Question: Which is faster e4x or object? Has anyone really benchmarked this? It seems that when I return an http request as an Object my parse results are faster than trying to do a similar parse in e4x. This seems to be particularly true when the XML has structure than can be returned as an ArrayCollection vs assigning the same xml via e4x to an XMLListCollection. I'm curious as to what result format most people are using for their HttpRequest. Thanks! Wayne
