[GHC] #3693: Show stack traces

GHC ghc-devs at haskell.org
Tue Aug 27 15:05:35 UTC 2013


#3693: Show stack traces
-------------------------------------+------------------------------------
        Reporter:  jpet              |            Owner:
            Type:  feature request   |           Status:  new
        Priority:  normal            |        Milestone:  7.6.2
       Component:  Runtime System    |          Version:  6.10.4
      Resolution:                    |         Keywords:
Operating System:  Unknown/Multiple  |     Architecture:  Unknown/Multiple
 Type of failure:  None/Unknown      |       Difficulty:  Unknown
       Test Case:                    |       Blocked By:
        Blocking:                    |  Related Tickets:
-------------------------------------+------------------------------------

Comment (by Tarrasch):

 Hello, I'm planning to take a serious look at this, it'll be my master
 thesis. I have a few questions here that I believe is of general interest.

 (1) Peter, your progress looks fantastic, and it's great to see that you
 can retrieve the stack trace both through a debugger ''and'' through the
 run-time system. But which one should we aim to use in the end? Opinions?
 I personally think having the run-time doing the job to be simpler for the
 users (If they are unfamiliar with debuggers) and more convenient (what if
 I forgot to run my program with `gdb`?)

 (2) By now, GHC have had the LLVM backend for a while. Would stack traces
 working for this backend be more useful? Would it be more/less work? LLVM
 has support for DWARF, but at this time I have a bad idea what that even
 means. But I do know some complications with the llvm approach. I've
 spoken with Nathan Howell about this and he thought that one missing piece
 is that the LLVM architecture only understands DWARF 3, but DWARF 4 is
 necessary to implement what we want. So we would need to patch LLVM.

 (3) Peter, would it be possible to show in somewhat more detail on how to
 reproduce the stack traces you pasted previously? It would be really
 useful if I could have some trail to follow when playing around.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/3693#comment:47>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler




More information about the ghc-tickets mailing list