[kepler-dev] [Bug 1334] - need installer for kepler

[Bertram Ludaescher]

Chris:

these are fantastic tips -- IMHO we should heavily leverage/build upon
the Ptolemy II way of doing things for the Kepler release!

Bertram


>>>>> "b" == bugzilla-daemon  <bugzilla-daemon at ecoinformatics.org> writes:
b> 
b> http://bugzilla.ecoinformatics.org/show_bug.cgi?id=1334
b> ------- Additional Comments From cxh at eecs.berkeley.edu  2004-02-09 10:41 -------
b> By way of introduction, I'm the release engineer for the Ptolemy project.
b> 
b> We are currently using ZeroG's InstallAnywhere.  It works reasonably well.
b> I did use InstallShield for the Ptolemy II 1.0 release in 2001.
b> I had a number of problems at that time with InstallShield, see
b> http://ptolemy.eecs.berkeley.edu/ptolemyII/ptII1.0/installshield.htm#bugs
b> maybe InstallSheilds has fixed these problems, maybe not.
b> 
b> In any case I recommend InstallAnywhere.  It is not cheap, we ended
b> up upgrading to the Enterprise edition ($3000? for initial purchase,
b> $1500? for annual support for the second and subsequent years?)
b> If there is a free installer available, I would consider moving away from
b> InstallAnywhere.
b> 
b> 
b> It turns out that a fairly powerful release mechanism is to build a
b> Web Start installer first, which will help sort out jar file contents issues,
b> and then use InstallAnywhere.  I found the build/test cycle to be faster with
b> Web Start.  $PTII/mk/jnlp.mk has rules for building a Web Start release.
b> See also $PTII/doc/webStartHelp.htm, which can be found on the web as
b> http://ptolemy.eecs.berkeley.edu/ptolemyII/ptII3.0/ptII/doc/webStartHelp.htm
b> 
b> There is also the applet code generator, which can be used to create applets
b> of Ptolemy models.  I was playing around with having the applet code generator
b> create Web Start installers, but have not quite finished.  
b> 
b> One can also reduce the size of the download by using tree shaking, where
b> we run the model, note which jar files are actually used, and ship only those
b> jar files.  Of course, we need to be sure to include the error handling
b> mechanism etc., so treeshaking works best for non-GUI intensive products
b> like applets that have a plot as an output etc.
b> 
b> Anyway, I can provide some help in this area.
b> -Christopher
b> _______________________________________________
b> kepler-dev mailing list
b> kepler-dev at ecoinformatics.org
b> http://www.ecoinformatics.org/mailman/listinfo/kepler-dev