[GHC] #14299: GHCi for GHC 8.2.1 crashed with simple function?

[GHC]

#14299: GHCi for GHC 8.2.1 crashed with simple function?
-------------------------------+--------------------------------------
        Reporter:  mathiassm   |                Owner:  (none)
            Type:  bug         |               Status:  new
        Priority:  normal      |            Milestone:
       Component:  GHCi        |              Version:  8.2.1
      Resolution:              |             Keywords:
Operating System:  MacOS X     |         Architecture:  x86_64 (amd64)
 Type of failure:  GHCi crash  |            Test Case:
      Blocked By:              |             Blocking:
 Related Tickets:              |  Differential Rev(s):
       Wiki Page:              |
-------------------------------+--------------------------------------

Comment (by bgamari):

 Wow, what a disaster. Thanks for reporting this.

 There are a few issues intertwined here. Let's cover them in turn.

 == Problem 1: Unexpected binding shadowing

 The first problem here is a bit subtle and has to do with the fact that
 GHCi accepts two different types of syntax for defining bindings. You can
 say,
 {{{#!hs
 let f n = n * f (n-1)
 }}}
 or you can use
 {{{#!hs
 f n = n * f (n-1)
 }}}

 In your case you used both; GHCi interpreted this as two distinct
 bindings, one shadowing the other. That is to say, first you defined `f 0
 = 0`, then you "overwrote" `f` with, `f n = n * f (n-1)`. When you end up
 with is a factorial function that doesn't define its base case and
 consequently will never terminate. When I do this on my machine I get the
 error,
 {{{
 $ ghci
 GHCi, version 8.2.1: http://www.haskell.org/ghc/  :? for help
 Prelude> let f 0 = 0
 Prelude> f n = n * f (n-1)
 Prelude> f 0
 Stopped in <exception thrown>, <unknown>
 _exception :: e = _
 [<unknown>] λ> :force _exception
 _exception = GHC.Exception.SomeException
                (GHC.IO.Exception.SomeAsyncException
                   GHC.IO.Exception.StackOverflow)
 }}}
 That is, a `StackOverflow` exception.

 In contrast, if I defining these two bindings in one GHCi command, I get
 the expected behavior (modulo the incorrect definition of factorial),
 {{{
 Prelude> f 0 = 0; f n = n * f (n-1)
 Prelude> f 0
 0
 Prelude> f 2
 0
 }}}
 or, using the `let` syntax,
 {{{
 Prelude> :{
 Prelude| let f 0 = 0
 Prelude|     f n = n * f (n-1)
 Prelude| :}
 Prelude> f 0
 0
 Prelude> f 4
 0
 }}}

 I can see that the fact that the syntax works out this way would be quite
 surprising. I'm not entirely sure what to do about it, however. Out of
 curiosity, what instructions are you following?


 == Problem 2: MVar exception

 This is almost certainly a bug in GHC(i)
 {{{
 ghc: panic! (the 'impossible' happened)
   (GHC version 8.2.1 for x86_64-apple-darwin):
         thread blocked indefinitely in an MVar operation
 }}}
 Unfortunately I'm able to reproduce on neither my Linux box nor an OS X
 machine, both running 8.2.1. Are you able to reliably reproduce this
 crash?

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14299#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler