IRC Logs for #circuits-dev Tuesday, 2013-02-12

*** Osso has joined #circuits-dev01:37
mehereprologic, when I tried it before, I only removed that if statement. You did a much more thorough patch. With you patch, nothing fails here when I apply it, so I think you should commit it.01:50
prologicexcellent02:21
prologicI have and pushed it up02:21
prologiccool :)02:21
prologicSo we're happy with these semantics?02:21
*** Osso has quit IRC03:18
*** Osso has joined #circuits-dev04:14
mehereYou didn't push it yet, right? Well, don't. I just noticed that while all unit tests run, the change breaks my application. Looking into it...04:47
*** mehere has quit IRC07:04
*** mehere has joined #circuits-dev07:04
mehereTricked ourselves out. See that07:12
meherehandler_channel = _handler.channel or getattr(07:12
mehere+                getattr(07:12
mehere+                    _handler, "im_self", getattr(07:12
mehere+                        _handler, "__self__", _dummy07:12
mehere+                    )07:12
mehere+                ),07:12
mehere+                "channel", None07:12
mehere+            )07:12
mehereWell, if handler.channel is a Component, then it is also Manager. Which means that using it in a boolean expression will call Manager.__len__ which happens to be 0 and not at all what we want here ;-)07:13
prologichmm08:28
prologicrhttp://pastebin.com/jfZryJJW08:34
prologicI think this should fix that.08:34
prologicWHat do you think?08:34
prologicHmmm08:39
prologicThis is wrong.08:39
prologicchannel can legitimately be None08:39
prologichttp://pastebin.com/m6JL8Wj708:46
prologicmehere, see paste above08:46
mehereDon't think so. This could effectively assign a bool to the handler. How about08:50
meherehttp://pastebin.com/9QQDhTjG It does make two tests fail, however, just looking at it...08:51
*** Osso has quit IRC09:00
mehere... and there are "hidden" problems (only visible in log), looking...09:06
*** anunnaki has quit IRC09:29
*** anunnaki has joined #circuits-dev09:31
*** anunnaki has quit IRC09:40
*** anunnaki has joined #circuits-dev09:44
mehereOK http://pastebin.com/kaZcVSr4 is what I am at by now. test_node.py still fails, but I first have to understand what the purpose of this test actually is.09:53
*** anunnaki has quit IRC09:55
*** anunnaki has joined #circuits-dev09:56
*** anunnaki has quit IRC10:15
*** anunnaki has joined #circuits-dev10:17
*** anunnaki has quit IRC11:59
prologicmehere, I pushed what I think is the fix12:58
prologiclet me know12:58
prologicI'll be at work in 2.5hrs12:58
*** anunnaki has joined #circuits-dev13:50
meherePushed all required fixes in individual commits for easier review. Will probably be sleeping when you arrive at work.14:09
meheremultipart formed worked here for both py26 and py32, but doesn't on shining panda. I think the implementation doesn't properly handle the fact that str in 2.6 depends on platform default (mine is utf-8, I assume panda's isn't).15:13
prologicmehere:  saw your changes. will review them a bit later15:32
prologicnice work15:32
prologicI think we've improve the semantics somewhat15:32
prologicI'll look into why test_node is failing15:32
prologiccircuits.node really needs more work ihmo15:32
prologicmehere:  agreed - I think we need to work out how to modify any unicode related code branches to be platform independent15:33
*** anunnakiii has joined #circuits-dev16:03
*** anunnaki has quit IRC16:04
meheretest_node works! Currently it's only the multipart form that fails (and that has nothing to do with our component-channel changes)23:54

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!