Re: [Haskell-cafe] Ambiguous type variable ‘f0’ arising from a use of ‘f’

On Sun, Mar 20, 2016 at 05:36:42PM -0400, Richard Eisenberg wrote:

On Mar 20, 2016, at 4:33 PM, Tom Ellis wrote:

...

According to Travis the old program compiled on 7.8 and 7.10. It certainly compiled on 7.6 on my local machine. When was the change that made it break introduced (and what was that change exactly)?

The change was for 8.0, as part of the redesign of part of the typechecker to support TypeApplications. The change in behavior that you're seeing here was not intentional, but was just a side effect of the redesign. It was originally reported as a regression, but I'm pretty sure that your original program should have been rejected, according to my understanding of how the monomorphism restriction works.

Oh, so you're saying it used to work by accident?