Actually, just an FYI: javac doesn't do anything different when you pass it the -O option anymore. They changed that as of JDK 1.3 (reference http://java.sun.com/j2se/1.3/docs/tooldocs/win32/javac.html). As of 1.3, Sun moved optimizations to the runtime (java) to be done with the JIT (just-in-time) optimizing. So with that, the two options of interest are passing -client (which is the default if neither is given) or -server to the call to the java executable. For which one to pass to your program, something like CFMX should be run using java -server (which it probably is?). The difference between the two is that the server option should do more aggressive optimization at the expense of a longer initial startup time among other things. Check out http://java.sun.com/docs/hotspot for more info on HotSpot.
Jeremy Babcock -------------- Programmer III UC Davis - Reprographics email [EMAIL PROTECTED] voice (530)757-3518 fax (530)753-1841 http://reprographics.ucdavis.edu -----Original Message----- From: Pete Freitag [mailto:[EMAIL PROTECTED]] Sent: Tuesday, September 17, 2002 10:37 AM To: CF-Talk Subject: RE: FW: Jsp Vs Cfm (CFMX) -- Test Code You can tell the Java Compiler (javac) to optimize the byte code (.class) it generates by giving it the -O flag. I would assume that the java compiler MX uses does optimize, but if it doesn't that may be worth looking into. Macromedia can focus on making the CFML->Java code more efficient, while the java compiler folks can focus on making the Java Source->Byte Code more efficient. Then JVM folks can focus on making bytecode execution on specific platforms more efficient. This is one of the benefits of CF moving to Java. _____________________________________________ Pete Freitag CTO, CFDEV.COM http://www.cfdev.com/ -----Original Message----- From: Dick Applebaum [mailto:[EMAIL PROTECTED]] Sent: Tuesday, September 17, 2002 12:33 PM To: CF-Talk Subject: Re: FW: Jsp Vs Cfm (CFMX) -- Test Code Ya'know, in the olden days of maimframes, and CoBOL, RPG, etc, they used to resolve this sort of problem with an "Optimizing" compiler. The Optimzer would perform another pass (either pre or post-compilation) and try to optimize the code --- especially subroutines and loops (and other high-potential code). It seems that something of that order would be useful here (especially going from an untyped language to a strongly-typed language). The Optimizer could look for key words (for, do, etc,) and make some intelligent guesses about the variables involved, or better yet, substitute an optimized, precompiled, code block for the user-specified code. Again, I don't know much Java, but it seems that CF ought to be able to generate something much closer to the jsp code. Maybe this is an opportunity -- CFMX plugin anyone? Dick On Tuesday, September 17, 2002, at 09:06 AM, Dick Applebaum wrote: > here's the critical piece of gen'd Java code: > > STIME.set(coldfusion.runtime.Cast._Object(this.GetTickCount())); > for > (X.set(((java.lang.Object)("1")));_compare(this._autoscalarize(X),1000 > 0 > 0 > 0.0)<=0;X.set(coldfusion.runtime.Cast._Object((coldfusion.runtime.Cast. > _ > double(this._autoscalarize(X)))+(1.0)))){{ > > Z.set(coldfusion.runtime.Cast._Object((coldfusion.runtime.Cast._double > ( > t > his._autoscalarize(Z)))+(coldfusion.runtime.Cast._double(this._autoscal > a > rize(X))))); > } > } > ETIME.set(coldfusion.runtime.Cast._Object(this.GetTickCount())); > > Kind of Ugly! > > Dick > > ______________________________________________________________________ Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Archives: http://www.mail-archive.com/cf-talk@houseoffusion.com/ Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists
