URL.encode, which we use heavily, is significantly slower -- see bug http://developer.java.sun.com/developer/bugParade/bugs/4725737.html for details.
My test (100 threds) runs in: 1.3.1: 7.8 s 1.4.0: 102.0 s 1.4.1: 92.1 s 1.4.2: 37.9 s
... we may need to write our own until this is resolved.
Anyway, you've made me change my mind. +1 to drop 1.3 support. Rationale: at the current project stage, making the life of JMeter developers easy is a priority.
Salut,
Jordi.
[EMAIL PROTECTED] wrote:
I think we need to list the advantages to moving to JDK 1.4. As I see it, they are:
1. Potential for higher IO performance (use nio)
2. Potential for better memory management (using memory mapped files could alleviate some memory problems)
3. Faster regular expressions: Sun's regex implementation is entirely faster than Oro's. This, however, is a very minor point.
4. Ends the need to support SSL with jdk1.3 with separate JSSE libs
5. New GUI components can be used freely (ie date switcher or whatever it was)
Please feel free to add to the list.
Another possibility has to do with playing with code versions. Two possibilities:
1. Split off a JDK1.3 branch in cvs that supports JDK1.3 (I can't imagine this working)
2. Make releases only of ApacheJMeter_java.jar that people can use to upgrade v1.9.x (ie, they keep the 1.9.x core, but upgrade the Java Sampler jar whose code remains compatible with JDK1.3). The rest of JMeter is free to use JDK1.4 only features.
Regarding #2, there is already a 1.9 branch in CVS. If a new feature in the Java Sampler code necessitates an upgrade of the core, those changes can always be made in that branch and a new 1.9.x release made. In this way, we might all get what we want (Well, except it really doesn't solve the Jordi-laziness factor ;-) heh, relax, I'm way lazier than you).
-Mike
On 26 Aug 2003 at 0:48, Jordi Salvat i Alabart wrote:
Jeremy Arnold wrote:
Regarding #1: [...]
Or maybe that's more trouble then it's worth.
I guess so :-)
As you mentioned, perhaps consulting jmeter-user wouldn't be a bad idea.
I will, now.
Regarding #2: In the recent past, I don't think I've logged on to a system that doesn't have a 1.4 JDK.
I have. Actually, I seldom log into a system which does have it installed. That's probably because the ATG's products currently require 1.3.
[...] If we had to run on just any person's system I could see this as being a problem, but since JMeter is (presumably) being run mainly by Java developers, I would expect that a recent JDK should be available.
*I*, for one, need to run it on customer's systems. They all have 1.3 installed (since they need it to run our products), but installing 1.4 would be one extra step -- not a long one, but still...
Anyway, I agree that my lazyness is not an argument, so let's stick to whatever issue #1 recommends.
-- Salut,
Jordi.
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
-- Michael Stover [EMAIL PROTECTED] Yahoo IM: mstover_ya ICQ: 152975688 AIM: mstover777
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
-- Salut,
Jordi.
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
