Repository navigation
Conversation
081bda7 to
94ca230
Compare
ynishiza
changed the title
94ca230 to
83eb097
Compare
| data BaseMonad | ||
| = forall m. (Typeable m, Monad m) => LazyBaseMonad (forall a. m a -> IO a) | ||
| | forall m. (Typeable m, Monad m) => StrictBaseMonad (forall a. m a -> IO a) | ||
| = forall m. (Typeable m, Monad m) => BaseMonad (forall a. m a -> IO a) |
There was a problem hiding this comment.
I simplified this to just BaseMonad since there are different nuances of strictness which can get complicated to categorize.
Instead, if there is a difference in the behavior depending on the underlying monad, I opted to just check the actual monad type directly and explain with a comment, as done in one of the tests here.
Please let me know if you have any thoughts.
There was a problem hiding this comment.
Just the strictness assertions moved out of the writer tests
83eb097 to
1781938
Compare
L0neGamer
left a comment
L0neGamer
left a comment
There was a problem hiding this comment.
Pretty darn good! Very happy with what you've done here.
| main = do | ||
| ComposeT.test | ||
| WriterStrictness.test | ||
| defaultMain $ testGroup "Transformers" [ |
There was a problem hiding this comment.
possibly "transformers strictness"
| testGroup "evalStateT" [ | ||
| testProperty "Lazy" $ \(F1Bot (x :: () -> (Int, ()))) -> | ||
| let result = flip Lazy.evalStateT () $ Lazy.StateT (MkSolo . x) | ||
| in isLazy result .&. isStrictIn (x ()) (getSolo result), | ||
| testProperty "Strict" $ \(F1Bot (x :: () -> (Int, ()))) -> | ||
| let result = flip Strict.evalStateT () $ Strict.StateT (MkSolo . x) | ||
| in isLazy result .&. isStrictIn (x ()) (getSolo result) | ||
| ], | ||
| testGroup "execStateT" [ | ||
| -- NOTE: See note on evalStateT. | ||
| testProperty "Lazy" $ \(F1Bot (x :: () -> (Int, ()))) -> | ||
| let result = flip Lazy.execStateT () $ Lazy.StateT (MkSolo . x) | ||
| in isLazy result .&. isStrictIn (x ()) (getSolo result), | ||
| testProperty "Strict" $ \(F1Bot (x :: () -> (Int, ()))) -> | ||
| let result = flip Strict.execStateT () $ Strict.StateT (MkSolo . x) | ||
| in isLazy result .&. isStrictIn (x ()) (getSolo result) | ||
| ], |
There was a problem hiding this comment.
with all of these tests, does the underlying monad type not change anything?
There was a problem hiding this comment.
Thanks for asking, and that's always a difficult question..
I think at least one basic case is the strict data type which will almost always have slightly different behavior:
ghci> import qualified Control.Monad.Trans.State.Lazy as L
ghci> data A a = A !a deriving (Functor, Show)
ghci> data B a = B a deriving (Functor, Show)
ghci> L.execStateT (L.StateT @() (\_ -> A undefined)) () `seq` ()
*** Exception: Prelude.undefined
CallStack (from HasCallStack):
undefined, called at <interactive>:63:38 in interactive:Ghci21
ghci> L.execStateT (L.StateT @() (\_ -> B undefined)) () `seq` ()
()
It's a bit similar to the difference that was discussed here: #123 (comment)
Or did you have something else in mind..?
Ideally, we would like to be able to test for all possible cases and each of their behaviors, but comes at the cost of greatly complicating the tests... which is why I've so far opted on using Solo as described in the Writer test documentation.
|
I'll try give another review over the week. I couldn't see much wrong so it'll likely be an approval! |
L0neGamer
left a comment
L0neGamer
left a comment
There was a problem hiding this comment.
There is no more I can say about this test. thank you for it!
Tests for the Lazy|Strict State transformer.
The approach is analogous to the Writer tests in #127.
evalStateT,execStateT,mapStateT,withStateT,put,modify,modify',modifyMgetsfmap,<*>,liftA2,>>=(same as the Writer tests)liftListen,liftPass, since the behavior differs for these, as it did also in the Writer transformers