Actually that's what's I addressed later in the post: although both run as
pure java it's the differences in the implementation that would cause the
performance differences.  I'll also agree that the savings CF offers at the
development level could very easily translate to decreased performance at
the Java level.

It definitely opens the door to arguments, but not, by itself, to the "10
times faster" argument.  ;^)

I just don't like to hear any "free floating stats" thrown about.  If we had
some test code, some metrics, some. well. anything, that would be different.
But as it stands we're still just guessing.

I still think (nothing more than a guess!) that for an average application
there's just no way that JSP is 10 times faster on the some J2EE engine.
There's just to much variability in even small applications.

Jim Davis



From: Gaulin, Mark [mailto:[EMAIL PROTECTED]
Sent: Wednesday, September 08, 2004 8:55 AM
To: CF-Talk
Subject: RE: Java+JSP 10 times faster than CF? And Other Ruminations.

> CFMX runs almost identically to the way JSP runs: a high-level source file
> is compiled directly into Java Byte code - the file run on a request for
> both tools is pure, 100% no bones-about-it Java.  Considering that it's
hard
> to see how JSP could be definitively "10 times faster".

I'm not one to worry too much about raw language performance, but I don't
believe the implication of your statement that since both JSP and CFMX
compile to java byte code then they must both perform the same in every
respect.  For example, I would bet that CFMX is using object "wrappers" to
represent variables on the CF page, something like a class called
CfmlVariable.  This would be needed to support the loose typing system of
CFML.  That means that you could code the exact same algorithm for something
numerically heavy (for example) in JSP and CFML and the two versions would
be translated to very different java (and so to different java byte code).
This isn't always a bad thing, and it's something you could probably ignore
most of the time, but it's there and that opens the door to "10 times
faster" arguments.
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings] [Donations and Support]

Reply via email to