Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 347449

Summary: Black box tests
Product: [Modeling] AMP Reporter: Jonas Ruttimann <jonas.ruettimann>
Component: GeneralAssignee: Miles Parker <milesparker>
Status: NEW --- QA Contact:
Severity: normal    
Priority: P3    
Version: 0.9.0   
Target Milestone: ---   
Hardware: PC   
OS: Windows 7   
Whiteboard:

Description Jonas Ruttimann CLA 2011-05-27 09:41:59 EDT 
There are three example plugins:
org.eclipse.amp.amf.examples.ascape
org.eclipse.amp.amf.examples.escape
org.eclipse.amp.amf.examples.repast

Maybe we should add another plugin for SD examples.


Since we want all examples to compile and run error free at all times, I think it would be a good idea to reuse the MetaAbm models in those plugins for tests. For each and every MetaAbm model we could check if code generations works, if the Java compiler does not report any errors and if the execution of the simulation brings the expected results. Of course, these kind of test should be run automatically.

Good idea/bad idea??
Comment 1 Miles Parker CLA 2011-05-27 12:56:22 EDT 
Yes.. these would be more like integration tests then balck box test, I suppose. Anyway, please note that we already have all of the gen.tests. Those should be primary diagnostic tests and then the test for generation and proper execution of models below would act like a sanity check on all of that.

By the way, I *did* figure out a way to use SWTBot to auto-generate everything just as a user would. I have a test for creating a new mdoel and making sure that it generates, but I haven't included it in standard tests as I haven't been able to figure out how to get SWT Bot working on Hudson yet.

I think we should just deprecate the repast tests and examples at least for the mid-term -- repast has gone up a version and I'm not sure anyone cares about support for that. (IF you do and you're reading this bug, pipe up!)
Comment 2 Jonas Ruttimann CLA 2011-05-30 02:23:00 EDT 
Yes, integration tests would be the correct term I guess. Concerning repast I do not have any objections. We're working with Escape.

(In reply to comment #1)
> Yes.. these would be more like integration tests then balck box test, I
> suppose. Anyway, please note that we already have all of the gen.tests. Those
> should be primary diagnostic tests and then the test for generation and proper
> execution of models below would act like a sanity check on all of that.
> 
> By the way, I *did* figure out a way to use SWTBot to auto-generate everything
> just as a user would. I have a test for creating a new mdoel and making sure
> that it generates, but I haven't included it in standard tests as I haven't
> been able to figure out how to get SWT Bot working on Hudson yet.
> 
> I think we should just deprecate the repast tests and examples at least for the
> mid-term -- repast has gone up a version and I'm not sure anyone cares about
> support for that. (IF you do and you're reading this bug, pipe up!)
Comment 3 Miles Parker CLA 2011-05-31 14:39:49 EDT 
OK, no Repast. :) Let me know as soon as you have those examples, because it is really difficult for me to figure out what is going on with the model design (i.e. why the simulation start within an agent definition) without seeing how a model works.
Comment 4 Jonas Ruttimann CLA 2011-06-07 08:54:50 EDT 
Please see org.eclipse.amp.amf\examples\org.eclipse.amp.amf.examples.sd