ghc-mod + stack: Emacs hanging opening Haskell file

[bixuanzju]

ghc-mod + stack work fine from the command line, but using it inside Emacs hangs Emacs, buffer freezes on "Initialiing ...", and I have to C-g to cancel. Besides, if I use the cabal instead of stack, everything works fine. So I suspect there's something wrong in the elisp code for the stack integration.

I am using Emacs 24.5, both the elisp and binary are compiled from the github repo. I am happy to provide any other information, thanks!

[luigy]

hey I think I figured it out! stack logging of lockfile warnings messes up what the elisp code expects...
I removed the lockfile logging from stack and it doesn't hang anymore
cc @kazu-yamamoto

with vanilla stack:

....
CONTENTS + EOT
Failed to grab lock ("/Users/luigy/.stack/lockfile"); other stack instance running.  Waiting...
Lock acquired, proceeding.
Failed to grab lock ("/Users/luigy/.stack/lockfile"); other stack instance running.  Waiting...
Lock acquired, proceeding.
E: Failed to grab lock ("/Users/luigy/.stack/lockfile"); other stack instance running.  Waiting...
E: Lock acquired, proceeding.
Failed to grab lock ("/Users/luigy/.stack/lockfile"); other stack instance running.  Waiting...
Lock acquired, proceeding.
O: OK
% boot
% ghc-mod root
........

[bixuanzju]

@luigy It didn't work in my case. Where did you see that Failed to grab lock ... message?

[luigy]

add (setq ghc-debug t) to your dotspacemacs/config and make sure you don't have a dist folder around root dir

I'm also using a ghc-mod compiled from dev-stack branch ( I haven't tried it out with master yet) and latest elisp code that spacemacs installs for ghc-mod is fine

you should see a *GHC Debug* buffer after doing a C-g when it hangs on Initializing .... and most likely it should look similar to mine around the beginning

[bixuanzju]

@luigy Thanks, I can reproduce now. Wonder why there is other stack instance running when using ghc-mod inside Emacs.

[luigy]

@bixuanzju should be related to commercialhaskell/stack#872 ... stack path is called a few times in dev-stack to get different paths
get rid of that logging and stack support looks good so far

[DanielG]

Which branch are you guys even using? Anyways the lockfile logging stuff was fixed in stack master some time ago and I got a bunch of other stuff we needed merged, should go into stack-1.4, or so I've been told. Will be working on stack support again this Monday so just hang on for a while ;)

[luigy]

@DanielG nice! feels so good to have "stackReconfigure" back with stack + ghc-mod... working good so far 🙌 Haven't ran into any problems yet ;)

The lockfile warnings keep happening to me on stack master, though

[bixuanzju]

@DanielG I am always using the master branch, but lockfile problem still persists.

[DanielG]

@bixuanzju You're using stack from master too?

[bixuanzju]

@DanielG yes, everything from master

[DanielG]

Well you'll have to figure out where the other stack process(es) are coming from then since I can't reproduce this. It might have to do with ghc-mod legacy-interactive forking off a thread to build the find symbol table on startup. That has caused a race condition with cabal-helper before so it's not impossible that this is causing trouble again.

[ajnsit]

Can confirm. I had to modify the stack code to remove the code that prints the locking error message, and the problem got resolved!

[PierreR]

I can confirm this as well. By adding 2>/dev/null as mentioned in #454 I still get the message Failed to grab lock a couple of times (even with stack-1.4 version) but no more infinite loops.

As an aside, I have to use the ghc-mod binary generated by cabal in .cabal-sandbox. If I try to use one generated by stack install (in ~/.local/bin/), it hangs forever as well.

[bixuanzju]

@DanielG FYI, I am still having the same problem in the master branch

[DanielG]

That's all nice and well but without more information or a way to reproduce this I can't really do anything about it >_>

[kazu-yamamoto]

Now Daniel and I can reproduce this. Daniel is working on this.

[bixuanzju]

@DanielG awesome! The HEAD fixed my problem, finally I can embrace ghc-mod with stack!

[kazu-yamamoto]

This has been fixed on master.