review fix: launcher#setArgs cannot be called twice

[surli]

This PR shows that the current behaviour is to never reset the model, even if the same launcher is called twice with different arguments.

My current test shows that model and model2 are the same and they contain 2 types at the end of the test.

I don't know if it's an expected behaviour or not, but it looks very dangerous to me. WDYT?

[surli]

By the way, the only way to make this case working is by using setArgs() because then we call processArguments which create a new JDTBasedSpoonCompiler. If we try doing that using the API (addInputResource and so one), it will never work because we don't create back the JDTBasedSpoonCompiler.

Personally what I would want as a user is:

1. To obtain a new model each time I call buildModel()
2. To be able to call buildModel() several times for the same launcher

[pvojtechovsky]

I would say they spoon model is not ready to coexist with another spoon model made by same launcher. Note that each model produced by launcher actually MUST share Factory and StandardEnvironment and it causes at least problems with change event handling ...

So it looks like that idea needs deeper refactoring

[surli]

I would say they spoon model is not ready to coexist with another spoon model made by same launcher.

Ok then we should prevent calling twice setArgs on the same launcher by throwing an exception for example. Do you agree?

[pvojtechovsky]

we should prevent calling twice setArgs on the same launcher by throwing an exception

I agree ... and later we might think about that refactoring. It is not nice when model specific things are in StandartEnvironment...

[monperrus]

explicitly set it to false?

[monperrus]

add a comment about the meaning of the field?