search

ndmitchell / hoogle

Haskell API search engine
http://hoogle.haskell.org/
Other
738 stars 134 forks source link

Update expected test results for 8.6 base. #283

Closed matt-noonan closed 5 years ago

matt-noonan commented 5 years ago

This mostly involved some previously-expected-failures turning into successes, but there was a regression where Yesod's lookupJust :: (Eq a, Partial) => a -> [(a,b)] -> b is now getting preferred over base's lookup :: Eq a => a -> [(a,b)] -> Maybe b.

Fixes #282

ndmitchell commented 5 years ago

Thanks! Any idea why the new base made everything better?

matt-noonan commented 5 years ago

I'm not positive, but it looks like the big difference in the search results in that a bunch of happstack functions no longer appear. For example, this used to be a match for the query a -> [a]: ok :: (FilterMonad Response m) => a -> m a in happstack-server. I guess this will probably resurface and the tests will have to go back to being expected failures once these functions re-appear in the index.

ndmitchell commented 5 years ago

I wonder if we need to "weight" base higher in type searches? For string searches the base weighting is quite high.

matt-noonan commented 5 years ago

Indeed :) https://github.com/ndmitchell/hoogle/issues/267