jakarta-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebb <seb...@gmail.com>
Subject Re: [JMeter] Convert to Maven based build?
Date Fri, 26 Nov 2010 20:42:00 GMT
On 26 November 2010 16:54, Phil Steitz <phil@steitz.com> wrote:
> On Thu, Nov 25, 2010 at 4:29 PM, Rahul Akolkar <rahul.akolkar@gmail.com>wrote:
>> On Thu, Nov 25, 2010 at 3:44 PM, Tim O'Brien <tobrien@discursive.com>
>> wrote:
>> > On Nov 25, 2010, at 2:29 PM, Peter Lin <woolfel@gmail.com> wrote:
>> >
>> >> even though I haven't been active in jmeter in a while, I am still a
>> >> jmeter committer.
>> >>
>> >
>> > Quantify "a while".
>> >
>> <snip/>
>> No need, we have archives for the curious.
>> > I'm still a committee on various projects.   Would I veto a change by
>> someone with a defined need who shows initiative?   No.
>> >
>> > If Peter Lynch has the itch, why not let him experiment?   This place
>> works on initiative, not a series of subjective objections to a tool he
>> wishes to use.
>> >
>> > This place works only if people like yourself (like myself) get out of
>> the way of people more active than ourselves.
>> >
>> <snap/>
>> All good above.
>> Finally, the onus is on those who experiment to convince those who do
>> the work here that the proposed changes are then worthy.
>> +1
> As one data point for a potential contributor, I will share the following.
> I have been lurking on this list for quite some time and intending to
> eventually propose some ideas/patches for enhancements.

We look forward to seeing these!

> Seeing this thread,
> i thought it would be a good idea to see how hard it was for me to get set
> up to build the code and run the tests.  The answer is it took me about 10
> minutes, which is frankly less time than most maven-built projects take to
> get going and *way less* than any nontrivial maven build.  I particularly
> like that there is a README as well as a building.html that clearly describe
> the simple steps necessary to get set up.  If you follow the directions to
> download the dependent jars and replace the Eclipse .classpath file with
> eclipse.classpath, Eclipse is fully set up.  I did not try to actually run
> anything from within Eclipse, as I find that is in general a bad idea for
> anything nontrivial;

Eclipse can easily be used for running and debugging JMeter, but
running some of the JUnit tests can be a bit of a pain as they need
the classpath to be set up correctly.

> but the nicely documented ant build.xml worked for me
> out of the box.  It was impressive to me that I did not have to fuss with
> any local property settings, given the amount of config that Jmeter and its
> tests use.
> [I did get the following test failure:
> [java] 1)
> runSerialTest(org.apache.jmeter.junit.JMeterTest)junit.framework.AssertionFailedError:
> serialization of org.apache.jmeter.functions.gui.FunctionHelper failed:
> java.io.NotSerializableException: com.apple.laf.AquaComboBoxUI
> Looks Apple-specific.   I am running
>  java version "1.6.0_22"
> Java(TM) SE Runtime Environment (build 1.6.0_22-b04-307-10M3261)
> Java HotSpot(TM) 64-Bit Server VM (build 17.1-b03-307, mixed mode)]

It does seem to be Apple-specific, though perhaps the field that holds
the GUI item could be made transient to avoid the problem.
The JMeter GUI does not need to be serialisable - only the test
element etc. that are required for running in client-server mode.
The FunctionHelper is GUI only, so could be dropped entirely from the
serialisation tests.

> Two of the ten minutes were spent fussing with Eclipse because I had
> replaced the classpath before downloading the jars.  Closing and reopening
> the project was not enough to get Eclipse to stop thinking the jars were
> "missing."  I had to recreate it after the jars were in place.  It might be
> better to change the instructions to remind people to download the jars
> before creating the Eclipse project.   I can submit a patch for that if
> others agree this is a good idea.


> So I am personally finding it hard to believe that mavenizing the build is
> really going to make it easier for people to get involved with Jmeter.  If
> there are Jmeter artifacts of general usefulness, I think it would be a
> *good thing* to develop either Ant or Maven targets to get them published.
> That would be a much easier task than trying to get the full Jmeter build
> working in Maven.


Seems to me to be a useful gain.

Is that something that can be easily done in Maven, using the current
directory layout and jar contents?

Patches welcome (via Bugzilla please).

> I agree strongly with Rahul and Peter Lin though that this decision belongs
> to them who do the work.


I'm not against Maven per se, but I have not seen it used on a project
with so many output jars and which makes assumptions about locations
of jars.

I may be wrong, but I can see no benefit to converting to a Maven
build, and there are several potential blockers which I (and others)
have already mentioned else-thread.

I can see no point in starting the process when it might not be
possible to achieve its goal.

So I don't intend personally to spend any time working on creating a
Maven build for JMeter.

> Phil
>> -Rahul
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@jakarta.apache.org
>> For additional commands, e-mail: dev-help@jakarta.apache.org

To unsubscribe, e-mail: dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: dev-help@jakarta.apache.org

View raw message