[GHC] #11047: Provide call stacks in GHCi

[GHC]

#11047: Provide call stacks in GHCi
-------------------------------------+-------------------------------------
        Reporter:  simonmar          |                Owner:  simonmar
            Type:  task              |               Status:  new
        Priority:  high              |            Milestone:  8.0.1
       Component:  GHCi              |              Version:  7.11
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #545, #4837       |  Differential Rev(s):  Phab:D1407
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonmar):

 Yes - we would need to clearly define the interface between GHC and the
 interpreter as a set of commands that have to be serializable across the
 process boundary.  I imagine there would be a `mkByteCodeObject` command
 that took byte-code as an argument, and we would have to keep track of the
 byte-code with StablePtrs (or something like that).

 Luite has already done some of this work in order to get TH working with
 GHCJS - all the methods of the `Quasi` class are serialized using
 `Data.Binary` across the process boundary, including all of the TH syntax
 returned by a splice (see https://github.com/ghcjs/ghcjs/tree/master/lib
 /ghcjs-prim/GHCJS/Prim/TH).

 Regarding the semantic bugs, the main one is #5654.  I'll try to fix that,
 but it doesn't seem to happen all that often, and for a debugging tool I
 think an occasional inaccuracy isn't too much of a concern. Some
 information is better than none, and we get *lots* of good information.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11047#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler