FEST-478: Add support for parametrized types in AbstractAssert #11
Conversation
|
Thanks I hope to have time to review it this week |
|
Hi Mikhail, I have made some modifications to your commit, in order to constrain the selfType in AbstractAssert constructor, it is now : Class selfType instead of Class. I think this is better as it constrains selfType. You can review this in this branch or directly in this commit. Tell me what do you think and if it still fulfill your initial demand. Cheers, Joel |
|
Hi Joel, I looked at the code. I had this idea too. I thought this is not a good way of designing an API to make it's users have to deal with such warnings. But here we can't do anything better because of Java's type system. Now i understand that this approach is type-safer even thought it produces warnings. Anyway, it solves my original problem. Thanks! |
FEST-478: Add support for parametrized types in AbstractAssert
Mikhail,
Good point, we should offer the easiest API to our users, so letting them dealing with generics warning is not a good idea.
I thus prefer we deal with lesser type checking since it is mostly internal (I said mostly because when extending assertions the constructor now accept any class :
```java
protected AbstractAssert(A actual, Class<?> selfType) {
myself = (S) selfType.cast(this);
this.actual = actual;
info = new WritableAssertionInfo();
}
```
Anyway, when I use a third party library and I must do some cast due to generics I always feel bad, because I wonder if I did use the library the wrong way or if the library force me to cast.
So let's merge your work.
https://jira.codehaus.org/browse/FEST-478