Re: [boost] Any interest for stack_trace?

From: Jeff Garland (jeff_at_[hidden])
Date: 2007-04-20 19:21:22

Dirk Reiners wrote:
> 	Hi All,
> 
> I would be very interested, but having to instrument the code makes the use too 
> complicated IMHO.

I agree, that's what logging is for...course we don't have that lib either :-(

> There are some systems out there that do stack traces (I know VRJuggler does 
> them, and so does bugle, and probably many others), so there is probably code to 
> start with, if somebody wants to give it a try. The system-dependence and symbol 
> demangling makes it somewhat painful...
> 

The code I pointed to earlier uses non-standard API calls (others have pointed 
out similar facilities) to get a stack trace.  Ultimately at the next layer up 
what I want is something like:

//snaps stack state at time of construction
class FancyException : std::exception()
public:
    FancyException(const std::string& what);
    std::string backtrace()
}:

...
throw FancyException("Oh no");

...
catch (FancyException& e) {
   std::cerr << e.what() << std::endl;
   std::cerr << e.backtrace() << std::endl;
}

Or you could have a vector of strings for each stack element.  I think 
something like this could be done as a portable interface...it's just you'd 
get different information in the trace depending on how things are compiled, 
what platform, etc.

Jeff