Testing framework about ceylon-sdk HOT 9 CLOSED

Without annotations, the metamodel and anonymous functions it's going to be rather rudimentary (especially without annotations).

from ceylon-sdk.

I agree. But its going to be rather hard to test the sdk without something.
We can retrofit that something to be better once we've got annotations.

from ceylon-sdk.

where should we put this code? it should be written in Ceylon of course, and it will be useful for sdk and also ceylonjs. The ceylon.language project already has some assert functions but to use them I have to copy them to the separate ceylonjs modules and it would be so much better to start using the exact same stuff. And that will force us to make module loading work in ceylonjs...

from ceylon-sdk.

I've made a start on something junit-like for the purposes of testing ceylon.math. So right now it's in a com.redhat package in ceylon-sdk (or will be once I push ;-)

Can we legitimately use it for ceylon.language? I mean, that's a little bit circular.

from ceylon-sdk.

mmmm you're right... I guess we'll stick with the simple asserts there.

from ceylon-sdk.

I've pushed the basic test harness I mentioned if anyone wants to take a look. Current drawbacks:

• it doesn't work in Eclipse, only from ant.
• The lack of annotations and reflection means you basically have to register the test methods by hand in the subclass of Suite (see MathSuite).

from ceylon-sdk.

I also have such a module myself, I'll push it to Herd.

from ceylon-sdk.

Now that the JVM compiler supports methods with defaulted parameters and type parameters it should be possible to improve the test harness (specifically the comparison function uses for asserts can now be something like boolean compare<T>(T expect, T got)).

from ceylon-sdk.

This was fixed by @thradec I believe.

from ceylon-sdk.