Add :info command in Eval plugin #1948
Conversation
|
|
||
| import InfoUtil (Eq, Ord, Foo) | ||
|
|
||
| -- >>> :info Foo |
There was a problem hiding this comment.
I ended up changing these tests to work on a custom Foo (and Bar) datatype defined in InfoUtil, since using Bool meant way too many instances and potentially broken tests when new instances get added
| @@ -0,0 +1,3 @@ | |||
| packages: | |||
There was a problem hiding this comment.
I moved the test/testdata/cabal.project file one directory up, so I could define another library info-util. This was because :info prints the origin comment (-- defined in "...") using the full path to the file for modules in the same package, which wouldn't be portable.
| import Prelude (Eq, Ord) | ||
|
|
||
| data Foo = Foo1 | Foo2 | ||
| deriving (Eq, Ord) |
There was a problem hiding this comment.
I ended up dropping the Show instances since they would still come out in different orders sometimes. Thankfully Eq, Ord and Baz are enough to show the difference between :info and :info!
| mb_stuffs <- mapM (GHC.getInfo allInfo) names | ||
| let filtered = filterOutChildren (\(t,_f,_ci,_fi,_sd) -> t) | ||
| (catMaybes mb_stuffs) | ||
| return $ vcat (intersperse (text "") $ map pprInfo filtered) |
There was a problem hiding this comment.
You could always sort the filtered children before printing to ensure that the output is deterministic
There was a problem hiding this comment.
yeah, the problem is the [ClsInst] part of the tuple returned by GHC.getInfo. Sadly, ClsInst doesn't have an Ord instance, and I didn't want to impose any custom ordering
|
hmm, seems like |
|
mmm maybe different files for ghc version (I hope only will be two)? |
| @@ -107,11 +108,56 @@ tests = | |||
| ] | |||
| , goldenWithEval "Works with NoImplicitPrelude" "TNoImplicitPrelude" "hs" | |||
| , goldenWithEval "Variable 'it' works" "TIt" "hs" | |||
|
|
|||
| , testGroup ":info command" | |||
| [ testCase ":info reports type, constructors and instances" $ do | |||
There was a problem hiding this comment.
As suggested by @jneira, these tests now only look for the relevant infix strings (or their absence)
maybe it worths open a ticket in ghc asking for make it public? |
|
@akrmn many thanks for this useful feature |
@akrmn i was about to open an issue in the ghc project but is not obvious for me what definitions from |
99% of the code is copied from
GHCi.UI.infoand the functions called from there. Ideally this would be exported fromGHCi.UIbut sadly that's not the case.Note that the tests commit is marked as [WIP] because, when running the tests on my machine, I keep getting the instances in different orders. I haven't been able to figure out the source of nondeterminism.
This should eventually close #402