review: fix: Source position of CtCatch and CtCatchVariable

[pvojtechovsky]

@tdurieux do you agree that the test added by this PR should pass?

CtTry tryStatement = ...;
assertEquals("catch (IOException e) {\n" + 
		"			throw new RuntimeException(e);\n" + 
		"		}", contentAtPosition(classContent, tryStatement.getCatchers().get(0).getPosition()));

note: actually CtCatch source position contains only from parameters. E.g.: IOException e, which is not correct.

WDYT?

I will try to fix it in next days if you agree.

[tdurieux]

Yes, I agreed.

to get the catch position you should look at the jdt parent element.
I kind of did it for the catch variables

[tdurieux]

It seems good to me. Thanks @pvojtechovsky

[pvojtechovsky]

I am done here.

[tdurieux]

I'm afraid of this exception because there is a chance that we forget to handle one specific case and it will break the construction of the model.
I agree it is good to fix the problem but do you need to force this behavior to the users.

[pvojtechovsky]

What do you suggest? To silently ignore our mistake ("we forget to handle one specific case") and to produce invalid source position?

[tdurieux]

I don't know what is the best solution.
I just thought you can use the environment enableCheck, this way the user can disable it and you still have a chance to have a feedback.
WDYT?

[tdurieux]

I meant skipSelfChecks

[pvojtechovsky]

ok, I will do so

[pvojtechovsky]

PositionBuilder now writes into log and returns SourcePosition.NOPOSITION when there is a position detection problem and model checks are disabled. If they are enabled then it throws exception on each such problem.

[monperrus]

thanks Pavel