I'm frequently using the inlineCallbacks idiom for writing code, but am often finding myself frustrated with rather counterintuitive tracebacks. Let's say I'm writing an XML-RPC handler, and even if an exception occurs, I want it to return something to the user and dump the exception to the log file. However, if an exception occurs when I'm using the inlineCallbacks decorator, I sometimes get lacking tracebacks. Maybe I'll have the line number of the function that yielded the exception, but I'll have no detailed information as to where the exception really came from. For example, in the example I've pasted at the end of this mail, I get a traceback as follows: Traceback (most recent call last): File "exceptions.py", line 13, in xmlrpc_dosomething result = yield self._do_something_else() Exception: die die die!!! However, the exception happens within _do_something_else on line 20. I have a simple complete example copy/pasted below. Is there a better way to accomplish this? Thanks. - Paul Goins ----------------------------------- from twisted.web.xmlrpc import XMLRPC from twisted.web.server import Site from twisted.web.resource import Resource from twisted.internet import reactor, defer import sys, traceback class MyApp(XMLRPC): @defer.inlineCallbacks def xmlrpc_dosomething(self): try: result = yield self._do_something_else() defer.returnValue(result) except Exception: print >> sys.stderr, traceback.format_exc() @defer.inlineCallbacks def _do_something_else(self): yield raise Exception("die die die!!!") @defer.inlineCallbacks def cause_trouble(app): val = yield app.xmlrpc_dosomething() print val reactor.stop() if __name__ == "__main__": app = MyApp() r = Resource() r.putChild("RPC2", app) s = Site(r) reactor.listenTCP(8123, s) reactor.callLater(0.2, cause_trouble, app) reactor.run()