[kepler-dev] Subject: Kepler Startup times

Tony O'Hagan tohagan at itee.uq.edu.au
Wed Jun 6 17:54:32 PDT 2007


Chad,

I agree that my suggestion would not affect run-dev since it does not build
or load kepler.jar.

My observation was based on the fact that I noticed that ...
  kepler-1.0.0beta3/build/kepler.jar 
... contained a mixture of both .java and .class files and is easily the
largest jar file.  This is also the file that most first time *end users*
would load.

I also realise that the relative performance improvement of this change may
be minimal given other contributing factors but I figured that the
observation might still be useful.

Tony.

> -----Original Message-----
> From: Chad Berkley [mailto:berkley at nceas.ucsb.edu]
> Sent: Thursday, 7 June 2007 2:26 AM
> To: Tony O'Hagan
> Cc: kepler-dev at ecoinformatics.org
> Subject: Re: [kepler-dev] Subject: Kepler Startup times
> 
> There are two different tasks in the build file.  One is "run" which
> builds only jar files (I think with only classes in them, at least for
> kepler, not sure about ptII) and the other which is "run-dev" which runs
> kepler from the build/classes directory and doesn't build any jar files.
>   The run-dev takes less time to startup for developers who have to
> rebuild all the time and don't want to wait for jar files.  The run
> target probably takes less time to startup once the jar files are built.
>   I haven't tried comparing them though.
> 
> chad
> 
> Tony O'Hagan wrote:
> > Dan,
> >
> > One thing that you might consider to improve start up times is to build
> > separate jar files for .class and .java files.  The small jars
> containing
> > only .class files should load faster and cache better.
> >
> > Regards,
> > Tony O'Hagan
> >
> >
> >
> > _______________________________________________
> > Kepler-dev mailing list
> > Kepler-dev at ecoinformatics.org
> > http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-dev




More information about the Kepler-dev mailing list