Couple of things I forgot to mention: JMeter requires Java 1.4 as a minimum. Any add-ons need to support Java 1.4.
Also, it helps if the code has Javadoc for any new classes and data structures. No need to document methods that override standard JMeter methods, but it helps if new methods etc are documented. Unit tests are also very helpful. It's also useful to know if the add-on has been tested successfully in non-GUI and remote mode. S. On 10/03/07, sebb <[EMAIL PROTECTED]> wrote:
Sounds very useful. Bucket seems fine as a name to me. I seem to remember a similar idea being discussed before, but it never got as far as being built, so this is very apt. If you want to contribute the code, then the normal route (after checking on the mailing list as you have already done) would be to create a Bugzilla enhancement issue describing the contribution. You can then re-open the issue and attach any necessary files to it. Any modified files (e.g. properties) should be provided as unified diff patches against the current SVN (branches/rel-2-2). New files should not be provided as patches. Please make sure the files include the ASF license header at the beginning (just look at an existing file). It would help a lot if some text could be provided for the manual as well; this could either be a patch, or a <component> section in the style of the existing entries in component_reference.xml which could then just be pasted in. S. On 10/03/07, James Bull <[EMAIL PROTECTED]> wrote: > It's tentatively called Bucket listener which is probably not a good name. > > It provides information about the distribution of response times. > > You define a series of intervals and then it behaves in much the same way > as the summary element in the postprocessor menu only I've put it under > the listener menu. Each time a sample occurs it looks at the time and > increments the appropriate counter. I used a binary search to find the > correct interval so it shouldn't have too much effect on the amount of > load j-meter can generate even with a large number of intervals. It then > prints to the console window and the log file. > > You can specify the intervals manually in a table or you can just tell it > the upper and lower limits and the number of intervals you want and will > work them out itself. > > It uses the same property as the summariser to tell it when to report. > > I couldn't get it working as a separate jar file so I added the classes to > the apache core.jar > > The classes are > org.apache.jmeter.reporters.BucketListenerModel > org.apache.jmeter.reporters.BucketResultCollector > > org.apache.jmeter.visualizers.Bucket > org.apache.jmeter.visualizers.BucketSet > org.apache.jmeter.visualizers.BucketListener > > I would like to get this added to j-meter if possible. I want to know what > the procedure is for submitting this addition and what the standards for > acceptance are. I would also like to know if people think it is a good > idea and if they find it useful. > > My motivation in doing this was that I wanted to be able to run a large > number of samples (millions) against a website and have some knowledge of > response time distribution without having to store every result. I'm > reasonably pleased with it. I'm a tester rather than a programmer so my > code may well be a bit odd. I would be more than happy to receive > criticism of my code should someone have the time spare to have a look at > it. > > You can download the files from here if you are interested > http://www.box.net/p/james96046 > There are two files one is a built ApacheJmeter_core.jar with the bucket > listener in it and the other is just the source for it. > > James
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
