Proving App should make sure RL-Glue exists before eating their key (and RTS executable if RTS is the domain)

[GoogleCodeExporter]

I already had one extra key request because the person hadn't made RL-Glue,
and then the proving software ate their key and crashed when they did a run.

We can automated add logic to check that.

Original issue reported on code.google.com by brian.ta...@gmail.com on 3 Dec 2007 at 2:39

[GoogleCodeExporter]

Proving app now checks if RLGlue is built, havent yet checked about RTS.

Original comment by rad...@gmail.com on 3 Dec 2007 at 11:18

[GoogleCodeExporter]

Can you describe the new behavior? What does it do if RLGlue is not built?

Original comment by brian.ta...@gmail.com on 3 Dec 2007 at 11:20

[GoogleCodeExporter]

This is done for RL-Glue and just needs to be done for RTS.

Looking at the code, I think it could be refactored to be a 
processCreationFailed
exception, and then you could easily reuse a single chain of exceptions for both
missing RTS, missing RL_glue, and problems starting any process.  These are all 
good
situations to handle.

If we extend a runtime exception, we don't actually have to add "throws" 
everywhere,
as long as we can catch it later (need to check).

This is all very good, but I'm wondering if there is a more direct way to check 
if
the Glue is there before we actually start to fire everything up... ok I see 
how its
working.  Now I wonder why not just throw up the exception 1 more step, and 
then we
can just handle it in the big try-catch in runExperiment.  Maybe lets try that.

Also check out this page and see if we can use this type of exception to reduce
putting throws everywhere:
http://java.sun.com/j2se/1.4.2/docs/api/java/lang/RuntimeException.html

Original comment by brian.ta...@gmail.com on 7 Dec 2007 at 8:41

• Changed state: Started

[GoogleCodeExporter]

Ok, I actually acted on this for RL_glue because I wanted to end the night with 
some
fun refactoring.  Hope you're ok with it Matt.  Update and see if you feel its
cleaner. I like that now we don't have all sorts of guys who are not calling 
RL_glue
that still need to throw noRLGlueException.

Original comment by brian.ta...@gmail.com on 7 Dec 2007 at 9:00

[GoogleCodeExporter]

Added the similar check for RTS.

What's left is to print something like "There was a problem: 
"+Exception.toString() 
to the noRLGlueWindow.

And also it should probably be renamed to criticalErrorDialog or something.  
Anyway,
I Couldn't figure out how to write to it, so there is code to be written in the 
last
exception catch of runExperiment in Controller...

Original comment by brian.ta...@gmail.com on 7 Dec 2007 at 9:12

[GoogleCodeExporter]

All done and working for RTS and RL-Glue.

Original comment by brian.ta...@gmail.com on 8 Dec 2007 at 1:19

• Changed state: Fixed