[01:37] niemeyer: thanks for fixing gozk [01:38] niemeyer: 2012/06/05 11:37:12 JUJU provisioning: environWatcher reported error on Stop: watcher: critical session event: ZooKeeper connecting [01:38] now the PA can sense zookeeper going away [02:20] niemeyer: i'm liking the way this looks now [02:20] if we get a !ok watching for changes, we just return, and I let defer's take care of cleaning up [02:20] it's much cleaner, and -2 functions [07:59] what a crappy day [07:59] spent 1/2 the day thinking their was a bug in the watchers [08:00] then finaly realised, at 5pm, that there are actually connections in place when you are running tests [08:04] davecheney: Hiya. Hehe, I know that feeling. [08:05] davecheney: But at least you can end the day with the knowledge that you've found it. [08:05] and a rush to get my change propsed by 6pm :) [08:05] TheMue: i was going mad [08:05] i chased it all the way down to gozk and back again [08:06] and I was just writing up the message asking for help and explaining that if I actually killed the zookeeper server [08:06] then everything would work fine, when I relalised that clearly I was closing the wrong connection [10:09] hello. [11:25] Aram: Hi. [11:26] hey there [12:59] Goooood morning jujuers [13:00] What a beautiful morning it is here indeed [13:00] morning. [13:00] good morning. [13:00] rains here [13:00] here too [13:00] Aram: Worry not, I'll make sure to send some of the fantastic weather in your direction [13:11] niemeyer: do you see the transit of Venus in your area? [13:11] * Aram hopes it's not cloudy tomorrow [13:24] Aram: Unfortunately not :-( [13:39] niemeyer: Morning. Here it is grey, thakfully dry, and both daughters are ill. *sigh* [13:40] TheMue: Oh :( [13:40] TheMue: What's up with them? Cold? [13:41] niemeyer: No, the younger one has typical female probs (she always has a hard first day), the older one has cut herself yesterday. [13:41] Ouch! [13:42] niemeyer: Indeed, stayed until 1am in the clinic and hope to get her back today or tomorrow. [13:42] TheMue: Wow.. so it was serious indeed.. is she ok now? [13:43] niemeyer: Yes, but the cut has to be stitched. [13:44] TheMue: Ok, but that's generally something fast to do [13:45] niemeyer: Yes, now they only control that it doesn't infect. And her blood pressure went down. [13:48] TheMue: I see [13:49] niemeyer: That's why we now wait for a call from the hospital to fetch her. [13:50] niemeyer: By the way, could you take a look at bug 1007373? I added a comment what we already check and what to add when adding a relation. Maybe you see more. [13:50] TheMue: Will look [13:52] TheMue: It's not clear to me how your comment relates to the description of that bug [13:52] TheMue: Can you please respond (here or there) in terms of the specific issue described? [13:53] TheMue: THe issue is very specific.. a "Tests so far" list does not make it clear to me [13:53] niemeyer: OK, try to make it more clear. [13:53] TheMue: Talking to me is fine as well :) [13:54] niemeyer: The implemented tests so far are the same as in Py. That are all those points I listed below. [13:54] niemeyer: Those tests we not yet have - in both versions - are the open tests. [13:55] niemeyer: Mostly if identifiers are not empty and role and scope contain valid values. [13:55] TheMue: Can you please read the description of that bug again? [13:55] TheMue: And explain how you feel about *that specific issue* [13:55] TheMue: Your telling me about tests you have or not.. I raised a specific problem that I'd like to understand [13:56] niemeyer: As far as I understood the issue there are open validations regarding the endpoints in State.AddRelation(). [13:57] TheMue: Not just that.. it's also about the side effects of validating later rather than sooner [13:58] It'd be bad to be creating state in ZooKeeper just to later tell [13:58] something [13:58] > trivial to the user that we could have verified upfront. We shouldn't be [13:58] > duplicating this logic, though. [13:58] TheMue: I haven't heard anything about what you think of that yet.. [14:01] niemeyer: I'm just checking my current version, most is now tested upront any writing to ZK. [14:03] TheMue: Sorry, but I still can't follow.. can you please be more specific? Something like "No, that can't happen because if the endpoint is invalid foo bar will check and prevent the node from being created." [14:03] TheMue: You may well be right, but I can't tell yet.. [14:04] niemeyer: That's exactly what I'm implementing now and what the next proposal will show to you. [14:04] TheMue: Heh [14:04] niemeyer: I only wanted to know if you see any specific tests I forgot, so I wrote them down. [14:04] niemeyer: The way I test it you will see then. [14:05] TheMue: That bug is about one specific issue.. that comment makes no sense in that context [14:05] TheMue: We already have a function to validate endpoints.. [14:05] TheMue: That bug is about the fact they are not being validated before the endpoint is being acted upon, which causes state in zookeeper to be created. [14:05] niemeyer: Then maybe I've got a problem understand your issue and would like you to rephrase it for me. [14:06] TheMue: We need to do the validation upfront rather than doing random writing and then testing that the parameter is invalid. [14:06] TheMue: That's all [14:06] niemeyer: Not everything is tested yet, and when having two endpoints the first one could be writte while the second later breaks the iteration over the endpoints because the second one is invalid. [14:07] niemeyer: Yes, that upfront testing is added in the new branch. [14:07] TheMue: That's what this bug is about [14:07] TheMue: If you're fixing that, it's great [14:07] niemeyer: OK, then I only found more to validate. [14:07] TheMue: Yep, that's even better, thank you [14:07] niemeyer: The rest is clear. [14:08] TheMue: Awesome, cheers! [14:08] niemeyer: cheers [14:08] niemeyer: Then I only have to add some negative tests to verify it and it comes in. [14:40] * TheMue just got the news to fetch Janina in 2h from the hospital. *yay* [14:49] TheMue: Yay indeed [14:49] TheMue: Must have been scary [14:50] TheMue: Phew! [14:51] mramm: Yes, it has been. At least the first shock. After we've seen her it got better. But it's still different from a children's disease. [14:52] * TheMue thinks back of what we already had with our girls. [17:59] Woohay.. first fire of the year in the fireplace [19:55] cold and Brazil just doesn't compute for me for some reason [19:55] perhaps because I've never been there [19:55] or perhaps just because the heat is talked about far more than the cold [19:58] fwereade: ping [19:59] Beret: yeah, it's a common illusion to think that the whole country looks like Rio de Janeiro :) [19:59] Fernando shows me pictures of beautiful beaches and Andreas talks of the heat, and there's my impression [20:02] Beret: Andreas isn't a good parameter.. he probably has his own snow generation engine at home [20:04] hah [20:47] Beret: I live in Rio de Janeiro [20:48] :) [21:13] andrewsmedina: Just today my wife was saying she'd like to visit Rio.. [21:29] davecheney: Heya, [21:29] davecheney: Good morning [21:29] Just in time for me to step out :) [21:29] niemeyer: morning [21:29] May be back later for more reviewing.. [21:29] no worries [21:29] thanks for your work overnight [21:29] especially that schema one [21:30] i'll abandon that branch [21:30] niemeyer: one final thing, with the machines.String(), i agree just printing it's itoa id is right, but it will be a larger change to the tests. I'll resubmit the branch in a bit [21:34] davecheney: Sounds good [21:34] Back in a few hours for some reviewing [21:34] kk