eclipse-shell-developer

[EclipseShell-developer] Re: [Beanshell-users] EclipseShell: script Beanshell from Eclipse

[Werner S. (murphee) <wer...@ne...>]

Krau=DF wrote:
> One more thing occured to me that may be important. In the original pro=
ject, I used 'user libraries'=20
> to reference the drools and jcr-170 libraries in the project build clas=
spath rather than individual jar files.=20

> Correct me if I'm wrong, but as far as I understand the classpath setup=
 code, only source-folders and jar-files are
 > covered. Project dependencies or classpath variables come with a=20
different getEntryKind() response.

You're right, that should be added; the whole Classpath thing is a bit=20
tricky, so I left this at this stage for now... at least for the=20
InprocessBeanshell version.

The External mode for Beanshell should work, though, as I basically=20
steal the CLASSPATH settings from the Java project, and add them in bulk=20
to the "Start Beanshell" Launch Configuration. Do you see a Launch=20
Configuration by that name among your Launch Configurations? It's=20
CLASSPATH settings should reflect the ones of your project;

--=20
Blog @ http://jroller.com/page/murphee
EclipseShell @ http://eclipse-shell.sourceforge.net/

[EclipseShell-developer] Re: [Beanshell-users] EclipseShell: script Beanshell from Eclipse

[Werner S. (murphee) <wer...@ne...>]

Krau=DF wrote:
> I just set up a separate project to run another test (see attachment).=20
> Exlipcitely selected the project node before starting up the shell and =
simply evaluated the following commands in a region:
> 	test=3Dnew com.gft.test.beanshell.HelloWorld();
> 	test.setMyText("huhu");
> This time, I had a short glance at the error log as well. =20

BTW: for the external launch: it's good to evaluate "1+1" or something=20
like that first, to start the remote JVM. This takes a bit (a couple of=20
seconds), so just executing something simple is best; after the console=20
  for the new JVM process has appeared (it should have a red stop=20
button, like all other Java Consoles) see if the connection is up, do=20
something like print("Hello"); to see if it works.


> Using the 'external' run mode yields an rmi exception (see first two en=
tries in the logfile).

Yes, that should be about right, but they shouldn't be problems (I just=20
log them for completeness). The first one comes from the fact that I=20
want to execute something, but the newly launched JVM (with Beanshell)=20
hasn't managed to start the RMI server yet.
With the 2nd one, it tries to send a Beanshell Callstack over the wire,=20
which isn't possible; Did an Exception or an error occur in the=20
Beanshell JVM?

>It worked at least once (in a previous session) and I was able to issue =
a 'javap(test)' with the correct output,=20
 >but I can't reproduce it. I even rebooted the computer just to be on=20
the safe side, but couldn't get it working once more. Strange...

What happens now if you launch the external mode?


> For the eclipse mode, I get an entry in the error log referring to my w=
orkspace location.=20

But besides the error, is it possible to evaluate code?

> The problem is probably because of the use of the result of "path.toOSS=
tring();" in the InProcessBeanshellProvider.=20
 > Seems like the windows separator characters are interpreted as escape=20
charaters in the course of the evaluation of the 'cd' command.  >Maybe=20
you can get around this issue by setting the variable 'bsh.cwd' directly.

Thanks for the tip! The fact that the backslashes on windows cause=20
troubles, when the Beanshell interpreter sees them in Strings is really=20
a nasty little bug. I now set the bsh.cwd directory, and later on still=20
try the Beanshell cd();


Thanks for your bug reports, they're very helpful. Sorry if some things=20
don't work quite right yet, but with reports like these, I can try=20
fixing them.

murphee
--=20
Blog @ http://jroller.com/page/murphee
EclipseShell @ http://eclipse-shell.sourceforge.net/