On Sat, Apr 22, 2017 at 5:41 AM, Antony Polukhin via Boost <boost@lists.boost.org> wrote:
2017-04-11 4:20 GMT+03:00 Emil Dotchevski via Boost <boost@lists.boost.org>:
I have a question on stacktrace: is it possible to decouple the capture of stacktraces from the conversion to a human-readable format? These are two different use cases and while most programs need both, that is not true for libraries.
Done. Now if you include boost/stacktrace/stacktrace.hpp you would be able to catch stacktraces but would not be able to decode it. Now boost/stacktrace/stacktrace.hpp does not include windows.h, COM, LexicalCast and other "heavy" Boost libraries
Similarly, what I'd like is to grab a stacktrace, which is basically just addresses, then later match that up with, for example, pdb files, to get function names etc. When I push as task onto an async queue, I'd like to capture the stacktrace of the code doing the push, and include that with the task in the queue. So when the thread that performs the task throws or crashes, I can log "who" pushed that task. Then, back at work, decode that log and say "here, it's your fault". But the capture needs to be fast and small. Thus just addresses. Tony