Re: [PLAN] New Jasper
Henri Gomez wrote: > May be because of license of javacc ;-) That hasn't stopped Velocity from using it. But in any case, there was an announcement on the JavaCC list a couple of days ago that JavaCC will be open sourced over the next few months. Louis Tribble -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
Re: [TC4.0] Disabling debug output...
Craig wrote:
> - The typical approach with the JSR47 API or with Log4J is to let the
> logger decide whether to really output the message or not. So, you
> would instrument your code with things like (conceptual, not real API):
>
> logger.log("The object is " + myObject, Logger.DEBUG);
>
> and let the logger configuration decide whether you care about debug
> messages or not.
>
> - However, there is a performance problem with leaving this code in --
> you will be doing the string concatenation (and the method call)
> whether or not you want debug messages printed! If you are liberal
> about debug messages, especially inside frequently executed methods,
> this can have adverse impacts on performance (and increased garbage
> collection).
Often the toString() are much worse than the concatenations. After we realized
what a HUGE impact this had on performance, we added a bunch of overloadings
so that you could convert the above to the equivalent of:
logger.log(Logger.DEBUG, "The object is {0}", myObject);
where the formatting is only done once it's known that the category is enabled
(we used named categories instead of levels, but that's another discussion). I just
went and looked at log4j and java.util.logging and it looks like both would require
going through a wrapper class to get this feature.
The number of overloadings is a bit of a pain for the implementer once you
consider that you would often want to log boolean, int, and in some applications
float parameters, but not for the caller.
Louis Tribble
Re: Compiling JSP with debug info
"Craig R. McClanahan" wrote: > > Larry Isaacs wrote: > > > Hi all, > > > > We have a need for a global way to have Jasper compile the > > JSP servlet source with debug information. I would propose > > defining a system property such as: > > > > jasper.compile.debug > > > > which if set to "true" would turn on the inclusion of debug > > info. Any objections or alternative property names? > > > > Wouldn't you really want this option at the per-web-app level, rather than > global? While we were at it, you'd want similar options for turning > optimization on and off, or (more generally) a way to set compiler options for a > generic compiler. Getting Tomcat to use the right compiler command to work with our debugger has been a headache, both in preparing documentation and in providing support. And Tomcat 3.1 and 3.2 have some differences besides. I would love to see this configurable in server.xml both globally and per context, and at a more abstract level than an initialization parameter to org.apache.jasper.runtime.JspServlet. Ideally, in fact, the configuration value would be the compile command string, whose meaning everybody already understands, for example: "javac", "jikes", "/usr/java/bin/javac -g", or (in our case) "mjavac -g". [It may be desirable to support a separate compiler type parameter (which can usually be inferred from the compile command) to deal with differences (e.g. classpath handling) between 1.1 javac, 1.2 javac, and jikes.] Invoking javac in-process could be viewed as an under the hood optimization (if the command is "javac" with no path, invoke in-process) but it could be exposed explicitly. We're about to release an inexpensive Tomcat edition of our debugger: for now, in order to have the same instructions for Tomcat 3.1 and 3.2, and given that we don't want to embed Tomcat, I haven't thought of anything better than recommending replacing tomcat.bat/sh with a slightly modified one that we supply (or editing the java invocations by hand if it's been customized). Comments? Louis Tribble -- <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> Louis Tribble [EMAIL PROTECTED] Metamata, Inc. http://www.metamata.com Tools for serious Java developers. +1 510 796 0915 <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Re: JDK bug or am I crazy ?
[EMAIL PROTECTED] wrote:
>
> Hi,
>
> I spent many hours trying to fix a bug in tomcat3.3. I'm now down to
> DependManager, and I really have no idea what's happening.
>
> There is a method:
>
> boolean shouldReload() {}
>
> All return statements are:
> "return expired;", where expired is an object property ( no local
> variables - the class doesn't extend and it's not extended ).
>
> I do a println(expired) just before every "return expired;" and it prints
> "false".
>
> I do a println(expired) after the method returns - it's still "false".
>
> The only problem is that the returned value is "true" - even if the
> function returns a value that is false before and after the return
> statement. ( the variable is final, and there is only one place that
> sets it true, and that point is never reached ).
>
> This doesn't happen if I use IBM JDK1.3 - only with Blackdown 1.2.2 ( FCS,
> BETA ).
>
> VERY, VERY SCARRY !!!
Does disabling the JIT help any? ( -Djava.compiler= )
Louis
--
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Louis Tribble [EMAIL PROTECTED]
Metamata, Inc. http://www.metamata.com
Tools for serious Java developers. +1 510 796 0915
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Re: Recent Tomcat Download Stats
Louis Tribble wrote: > > Presumably when 3.2 goes final, there will be a surge. > > Also, 4.0-m5 (this weekend or so) will be the first Tomcat 4.0 > "milestone" (alpha) that has the connectors to run with Apache. > > "Craig R. McClanahan" wrote: > > PS: Tomcat 4.0-m4 just went over 10,000 downloads of the binary distribution > > (since 11/01/2000). Tomcat 3.2-b7 and 3.2-b8 have totalled 8,095 downloads > > (undoubtedly with some overlap) in roughly half that amount of time. In both > > cases, there are an additional unknown number of people keeping up via source > > distributions, nightly builds, and/or anonymous CVS access. Oops, sorry about that. Usually, I know the difference between Reply and Forward. Although it does deserve more prominent mention than a P.S. in another thread :-) Louis -- <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> Louis Tribble [EMAIL PROTECTED] Metamata, Inc. http://www.metamata.com Tools for serious Java developers. +1 510 796 0915 <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Recent Tomcat Download Stats
Presumably when 3.2 goes final, there will be a surge. Also, 4.0-m5 (this weekend or so) will be the first Tomcat 4.0 "milestone" (alpha) that has the connectors to run with Apache. "Craig R. McClanahan" wrote: > PS: Tomcat 4.0-m4 just went over 10,000 downloads of the binary distribution > (since 11/01/2000). Tomcat 3.2-b7 and 3.2-b8 have totalled 8,095 downloads > (undoubtedly with some overlap) in roughly half that amount of time. In both > cases, there are an additional unknown number of people keeping up via source > distributions, nightly builds, and/or anonymous CVS access. -- <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> Louis Tribble [EMAIL PROTECTED] Metamata, Inc. http://www.metamata.com Tools for serious Java developers. +1 510 796 0915 <><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
