C-u C-c C-t doesn't add type signature

[gracjan]

Redditor has a problem:

The correct type is displayed in the mini buffer (basically C-u doesn't do anything, and I just get the result of C-c C-t.) How can I debug this problem and get automatic type annotations working again? Thanks.

https://www.reddit.com/r/emacs/comments/2rbv11/haskell_mode_cu_cc_ct_doesnt_add_type_signature/

[PierreR]

Is there a workaround for this issue ? I thought M-t would insert the type signature but for me it just prints "Nothing to be done".

As a note, C-c C-t does give me a proper type.

[bergey]

@PierreR Is this still a problem with current HEAD?

If so, is C-c C-t bound to inferior-haskell-type or haskell-process-do-type? Can you provide an example that produces the error?

[PierreR]

@bergey I am using haskell-mode-20151126.1250 and yes it is still a problem.

c-c c-t is bound to ghc-show-type. The function that offers this feature comes from ghc-mod if I am not mistaken: ghc-insert-template-or-signature. Is there an equivalent not related to ghc-mod (it looks like ghc-mod is no so active these days) ?

Here is the ghc-mod issue : DanielG/ghc-mod#628

[bergey]

I believe the best-maintained equivalent now is haskell-process-do-type. It assumes there's already a connected ghci session, so I call haskell-process-load-file first (C-c C-l by default) and then haskell-process-do-type (which is the current default for C-c C-t).

I'll try to look into it, now that I know which function is being called. (In case it's fixable in hasell-mode; I can't help on the ghc-mod side.)

@gracjan Can you change the title of this issue to "ghc-show-type doesn't add type signature".

[geraldus]

Due to ghc-show-type is not a part of haskell-mode I'll close this, haskell-process-do-type (and haskell-mode-show-type-at) inserts type signature in presence of universal prefix arg.

[PierreR]

@geraldus haskell-process-do-type shows the signature in the minibuffer but does not insert anything for me ...

Should you re-open the issue or shall I create an new one ?

[gracjan]

Lets reopen this here.

[novakboskov]

With:
haskell-mode-20161027.139 and GNU Emacs 25.1.1
the issue is still here.

C-c C-t is boud to haskell-process-do-type and it works well but C-u C-c C-t does not even call haskell-process-do-type (when I instrument this function and type C-u C-c C-t it does not stop the debugger and prints C-u C-c C-t is undefined in minibuffer.)

It looks a bit strange to me since haskell-process-do-type declares (interactive "P") in its definition.

Is there some additional level of indirection when calling this function? Other functions declare (interactive "P") are called properly with prefix argument.

[PierreR]

FWIW, I think it would be OK for haskell-mode to stop supporting this leaving it to external tools such as intero (it does work there) or ghc-mod (last time I checked it didn't work there).

The worst is to keep pretending that this feature works ;-)