Getting types inside a function in ghci

Hello all, is there a way to determine types of symbols which are not toplevel but inside a function?

On 29 October 2013 22:36, martin
is there a way to determine types of symbols which are not toplevel but inside a function?
Not at present without transformation. There are two reasonably reliable approaches: Insert noisy holes/make intentional type errors: http://ghc.haskell.org/trac/ghc/wiki/Holes#Insertingdeliberatetypeerrors Implicit parameters: http://ghc.haskell.org/trac/ghc/wiki/Holes#Implicitparameters Holes will land in GHC to support this explicitly. I don't know what version of GHC has/will have this. Alternatively, you can use ghc-mod (https://github.com/kazu-yamamoto/ghc-mod) which, if you can get it to work, can tell you the types of sub-expressions. Lastly, the next version of the FP Complete IDE gives you types of sub-expressions just by clicking on them or selecting them out of the box.

Am 10/29/2013 11:06 PM, schrieb Christopher Done:
On 29 October 2013 22:36, martin
mailto:martin.drautzburg@web.de> wrote: is there a way to determine types of symbols which are not toplevel but inside a function?
Not at present without transformation. There are two reasonably reliable approaches:
Insert noisy holes/make intentional type errors: http://ghc.haskell.org/trac/ghc/wiki/Holes#Insertingdeliberatetypeerrors
Implicit parameters: http://ghc.haskell.org/trac/ghc/wiki/Holes#Implicitparameters
Holes will land in GHC to support this explicitly. I don't know what version of GHC has/will have this.
Alternatively, you can use ghc-mod (https://github.com/kazu-yamamoto/ghc-mod) which, if you can get it to work, can tell you the types of sub-expressions.
Lastly, the next version of the FP Complete IDE gives you types of sub-expressions just by clicking on them or selecting them out of the box.
Thanks, that helps. Related questions: how can I figure out the type (using holes) for the symbol left of the <- in a do expression? I mean other than: main = do x <- getChar::undefined return () Couldn't match type `undefined' with `IO Char' And how about plain lambdas. This works f = \x -> (x::undefined)+1 Couldn't match type `t' with `undefined' but what if x does not appear on the right side?

On the left of <- you could do do () <- someThing …. so that you'll get a type error against (). Or use scoped type variables: do (x::()) <- someThing

FWIW, I asked about this on stackoverflow a while back, and there were a
couple of good approaches, although nothing that completely achieves it
from ghci. I've used the ghci breakpoint trick a few times, although it's
doesn't give the "complete" type.
Here's the link:
http://stackoverflow.com/questions/15034392/find-inferred-type-for-local-fun...
On Tue, Oct 29, 2013 at 3:06 PM, Christopher Done
On 29 October 2013 22:36, martin
wrote: is there a way to determine types of symbols which are not toplevel but inside a function?
Not at present without transformation. There are two reasonably reliable approaches:
Insert noisy holes/make intentional type errors: http://ghc.haskell.org/trac/ghc/wiki/Holes#Insertingdeliberatetypeerrors
Implicit parameters: http://ghc.haskell.org/trac/ghc/wiki/Holes#Implicitparameters
Holes will land in GHC to support this explicitly. I don't know what version of GHC has/will have this.
Alternatively, you can use ghc-mod ( https://github.com/kazu-yamamoto/ghc-mod) which, if you can get it to work, can tell you the types of sub-expressions.
Lastly, the next version of the FP Complete IDE gives you types of sub-expressions just by clicking on them or selecting them out of the box.
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
participants (3)
-
Christopher Done
-
martin
-
Tom Murphy