stack traces when debugging GHC

Simon Peyton Jones simonpj at microsoft.com
Mon Apr 13 06:57:15 UTC 2015


|  I see you've added a reference buried in Debugging/Compiler.  Why don't
|  we just make traceStack the default?  Perhaps with a flag to disable
|  (or enable) the stack trace.

Sounds desirable to me. It'll only work with profiling, though, of course.

Would someone like to do this?  At least make a ticket?

Simon

|  -----Original Message-----
|  From: ghc-devs [mailto:ghc-devs-bounces at haskell.org] On Behalf Of Simon
|  Marlow
|  Sent: 11 April 2015 21:27
|  To: Richard Eisenberg; ghc-devs at haskell.org Devs
|  Subject: Re: stack traces when debugging GHC
|  
|  On 09/02/2015 02:51, Richard Eisenberg wrote:
|  > Hi devs,
|  >
|  > I've just made a miraculous discovery and I want to share!
|  >
|  > It has happened many times that I create some bizarre error in GHC
|  and have no clue where it comes from. In the past, I've ended up using
|  pprTrace to label all calls of the failing function, sometimes
|  recursively, until I get an idea of what's going on. This is terribly
|  painful. And then, I noticed that the function Debug.Trace.traceStack
|  (just in the `base` package, not part of GHC) will print out a stack
|  trace, when profiling is enabled. So, I tried building a profiled build
|  of GHC, with -DDEBUG and -fprof-auto (just by editing the "prof"
|  section of build.mk).
|  >
|  > And then, voila, ASSERTs now print stack traces! And, with a small
|  change in Outputable, I can get pprTrace to print stack traces, too!
|  Hooray!
|  
|  I should have advertised this more widely, thanks for pointing it out.
|  
|  > I'd love to put this on the wiki. Any clue where this might be found?
|  
|  I see you've added a reference buried in Debugging/Compiler.  Why don't
|  we just make traceStack the default?  Perhaps with a flag to disable
|  (or
|  enable) the stack trace.
|  
|  Cheers,
|  Simon
|  _______________________________________________
|  ghc-devs mailing list
|  ghc-devs at haskell.org
|  http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs


More information about the ghc-devs mailing list