Add cause assertions to ThrowableAssert

[assertj]

Add cause assertions to ThrowableAssert

[jcgay]

Hi,

I wrote a test yesterday where I could benefit from using hasCause(), so I've tried implementing it.
It does what I need, but I'm not sure of what you've planned for this. As hasCause() there is using equals, the other assertion you proposed, hasCauseOfClass() seems necessary too. I feel that in most cases, we don't have access to exact reference of a Throwable to make assertions with ?

I will write hasCauseOfClass() to play with it and prepare a PR.
Thanks,

[joel-costigliola]

Yep, I think you are right, we don't usually have access to the cause exception instance so let's just keep these assertions :

hasCauseOfClass(Class<? extends Throwable> eClass);
hasRootCauseOfClass(Class<? extends Throwable> eClass);

or keep these but rename them to :

hasCause(Class<? extends Throwable> eClass);
hasRootCause(Class<? extends Throwable> eClass);

What do you think ?
I tend to prefer the second option but no strong opinion on that.

[jcgay]

I prefer the second option too.

It seems that @twillouer (although I don't see his comment, I just received an email) has already done most of the work to match cause using class, so let's wait for it :)

[twillouer]

@jcgay no time for the moment to finish this, your commit (jcgay@88e9b96) seems almost finished :)

[joel-costigliola]

One question though: do you think that hasCause(Class<? extends Throwable> eClass) should support inheritance ?
For example, if cause is a NullPointerException would hasCause(RuntimeException.class) be successful or not ?

I would say it should but it may be confusing ... WDYT ?

We may avoid the confusion by renaming the assertion to hasCauseInstanceOf(Class<? extends Throwable> eClass) and keep hasCause(Class<? extends Throwable> eClass) to check the exact runtime type.

[jcgay]

I think it should support inheritance. Naming it hasCauseInstanceOf sounds good to me, it's consistent with isInstanceOf behavior.

Exposing another assertion to verify exact runtime type can be useful too but should we not call it something like hasCauseExactlyInstanceOf to keep the naming from AbstractAssert ?

[joel-costigliola]

Agree !

Let's go for: hasCauseInstanceOf and hasCauseExactlyInstanceOf

[joel-costigliola]

Thanks Jean Christophe, very nice work !
You also added some examples for assertj-examples, much appreciated.

[jcgay]

Yep I forgot the examples the other time :)
Thanks !