Mailman 3 August 2010 - Twisted

[Twisted-Python] inlineCallbacks yield tuple support
by Сергей Магафуров Aug. 6, 2010

Aug. 6, 2010

This example shows what i want: @defer.inlineCallbacks def create_audio_link(tel, addr1, addr2): try: connection1, connection2 = yield (tel.connect(addr1), tel.connect(addr2)) except ConnectionError, exc: ... some error work ... else: return tel.create_audiolink(connection1, connection2) At now we can do this with DeferredList but it is complex (compare with first example): @defer.inlineCallbacks def create_audio_link(tel, addr1, addr2): try: try: r = yield DeferredList([tel.connect(addr1), tel.connect(addr2)], fireOnOneErrback=1, consumeErrors=1) except FirstError, exc: raise exc.subFailure.value else: connection1, connection2 = tuple(_r for _s, _r in r) except ConnectionError, exc: ... some error work ... else: return tel.create_audiolink(connection1, connection2) Approximate solution (idea) is patch in defer.py: def _inlineCallbacks(result, g, deferred): ... skipped ... if isinstance(result, Deferred): # a deferred was yielded, get the result. ... skipped ... waiting[0] = True waiting[1] = None if isinstance(result, tuple): # a tuple was yielded, get the result. result = DeferredList(result, fireOnOneErrback=1, consumeErrors=1) def gotResult(r): if isinstance(r, failure.Failure): r = r.value.subFailure else: r = tuple(_r for _s, _r in r) if waiting[0]: waiting[0] = False waiting[1] = r else: _inlineCallbacks(r, g, deferred) result.addBoth(gotResult) if waiting[0]: # Haven't called back yet, set flag so that we get reinvoked # and return from the loop waiting[0] = False return deferred result = waiting[1] # Reset waiting to initial values for next loop. gotResult uses # waiting, but this isn't a problem because gotResult is only # executed once, and if it hasn't been executed yet, the return # branch above would have been taken. waiting[0] = True waiting[1] = None return deferred I post this as ticket but was redirected to mailing-list http://twistedmatrix.com/trac/ticket/4627

1 0

[Twisted-Python] Get trial to use the docstring?
by Nathan Aug. 6, 2010

Aug. 6, 2010

Is there already a way to get twisted to print out a test's docstring (like unittest does) instead of the name of the function in it's output? I've looked at trial's options and searched the list archives, but I'm not seeing anything... For example, here's what my tests look like when run by unittest: $ test/run.py -v ==> Starting Prerequisite Tests ... ok Supported OS? ... ok Supported version of Python? ... ok ... And here's what they look like when run by trial: $ trial test/run.py run Test00Prerequisites test_00banner ... [OK] test_01os ... [OK] ... ~ Nathan

4 4

[Twisted-Python] Cred question
by Sury Soni Aug. 6, 2010

Aug. 6, 2010

Hi, I am using Cred authentication system in my Nevow based web application. Admin user can see list of all users from admin interface of the application. Now, admin user wants to do following things: 1. Automatically login as other user 2. Induce logout to already logged in user. Q1. Since, admin user itself used cred authentication to login in first place, how do we do automatic login as different user programmatically or logout already logged in other user? I tried to redirect my application to /__login__/?username=otherusername&password=itspassword and it went through Cred system, but I believe, since previous session was active as well, it triggered logout. Q2. Does cred allow multiple login sessions in same browser? So, that admin user can maintain its own session and login into other user as well. Please advice. Regards, Sury

3 3

[Twisted-Python] service stops/twistd exits
by Thomas Drake Aug. 5, 2010

Aug. 5, 2010

Hello, I'm wondering if there is an established pattern for having an application running in twistd shut itself down in the absence of a running service? The idea is that if a service fails it calls self.stopService() and the application shuts itself down somehow (rather than having the service call reactor.stop). Ideally, I was hoping to do with without any specialized code within my service, ie subclass from a ManagedService class or using decorators. I supposed it's possible to write a monitoring service that checks to ensure my service is running, but I'd prefer a tighter coupling like having a FatalException bubble up from the failed service to the application, MultiService, or whatever *should* be calling reactor.stop(). Otherwise, if having service (eventually) stop the reactor is totally antithetical to how things should work in a twistd plugin, then I guess I'll need to rethink things. Thanks, Tad

2 1

[Twisted-Python] Configuring Twisted logging via plugins
by Valeriy Zamarayev Aug. 5, 2010

Aug. 5, 2010

Hello, all. There have been a few questions on this list about how to configure Twisted logging via plugins, as opposed to .tac files. And we now know about the #638 and #3534 tickets. Hopefully, they will be resolved soon. But before this happens, it seems to me there's no other way to handle this than to use a "monkey patching" techinque on Twisted (excluding patching the Twisted source code itself). We need to access the 'application' object when the server starts. We can do this when the 'twistd' code calls setServiceParent on the service returned by the plugin. So if we wrap setServiceParent, we can access the application object and do what we want. See the code: http://gist.github.com/505926 This works for me. I don't know if a better and less dangerous way to achieve the same exists. Maybe there is one? Best Regards, Valeriy

3 3

[Twisted-Python] OpenStack uses Twisted
by Alvin Wang Aug. 2, 2010

Aug. 2, 2010

FYI - I went to a presentation on Nova/OpenStack https://launchpad.net/nova They use Twisted and they are planning on taking out Tornado. You can run your own amazon using Twisted.

3 5

[Twisted-Python] Weekly Bug Summary
by exarkun＠twistedmatrix.com Aug. 1, 2010

Aug. 1, 2010

1 0