Random with range#1875
Conversation
|
Did you actually run all the specs? Looks like this change breaks the build |
Fixed. Now travis return |
There was a problem hiding this comment.
Isn't the first check now duplicate since if the last one matches, the first one has to match too by definition?
There was a problem hiding this comment.
I agree. But with this change I need also little modify one spec. Check out new version :)
|
Does this spec change still make it pass on MRI too? |
No, because MRI does not execute |
|
Then the spec is problematic and we can't spec it like this. I think we need to explicitly check here for a float type. You should probably check how MRI does this exact coercion here so we can match the behavior. |
|
MRI check each type and try generate random number from each type (click)
|
|
Much better! |
MRI:
Rubinius before patch:
Rubinius after patch