Inconsistency of CtCatchVariable#type and #multiTypes

[pvojtechovsky]

The fields CtCatchVariableImpl#type and CtCatchVariableImpl#multiTypes both represents same role CtRole.TYPE - I am OK with that. because both works with same field - type.

But it is actually implemented inconsistently

//set type to x
CtCatchVariableImpl.setType(x)
//set type to y
CtCatchVariableImpl.setMultiType([y])
//ok, the type is y
assertEquals([y], CtCatchVariableImpl.getMultiType())
//what should next line do? This is how it behaves now
assertEquals(x, CtCatchVariableImpl.getType())
//but it should be
assertEquals(y, CtCatchVariableImpl.getType())
//set type to a,b,c
CtCatchVariableImpl.setMultiType([a,b,c])
//ok, the type is a,b,c
assertEquals([a,b,c], CtCatchVariableImpl.getMultiType())
//what should next line do? This is how it behaves now
assertEquals(x, CtCatchVariableImpl.getType())
//but it should probably throw an exception "There are multiple types in catch. Use getMultiType()"

WDYT?

I need to solve this problem before #1582 can be implemented.

[monperrus]

//but it should be assertEquals(y,CtCatchVariableImpl.getType())

yes, this should be the result

//what should next line do? This is how it behaves now assertEquals(x,CtCatchVariableImpl.getType())

x is is bug. however, instead of an exception, which makes client code brittle, I would prefer to return the first element of the list, or even better the first common super type of a,b,c

[monperrus]

closed per #1668