For reference this is what they would look like. Appologies now if my filter sends something beyond the party chat logs.
Martin ---------- Forwarded message ---------- From: [EMAIL PROTECTED] <[EMAIL PROTECTED]> Date: 2008/12/9 Subject: Chat with [EMAIL PROTECTED] To: [EMAIL PROTECTED] These messages were sent while you were offline. 12:42 partychat4: [EMAIL PROTECTED] joined the chat [EMAIL PROTECTED] there we go [EMAIL PROTECTED] hello all :) ["rob"] hi rg3 :-) 12:43 [EMAIL PROTECTED] hi rob [EMAIL PROTECTED] i tried getting in with Pidgin, but ended up giving up and going with kopete ["rob"] Ah - I normally have the opposite problem 12:44 [EMAIL PROTECTED] really? did you do anything special with it ? Pidgin seems to be intercepting the partychat comands and interpreting them so it wouldnt play ["rob"] ah ["rob"] you have to put a space first [EMAIL PROTECTED] oh :) 12:45 [EMAIL PROTECTED] HI Robbie [EMAIL PROTECTED] right, ill try that now...would rather use pidgin as then i can use it everywhere :) [EMAIL PROTECTED] hi marnie 12:46 [EMAIL PROTECTED] joined the chat [EMAIL PROTECTED] yup that works, got a nice error that im already in here :) ["rob"] :-) ________________________________ 16 minutes 13:03 partychat4: [EMAIL PROTECTED] well i think its time for lunch, bbiab :) ________________________________ 53 minutes 13:56 partychat4: [EMAIL PROTECTED] left the chat [EMAIL PROTECTED] joined the chat ________________________________ 41 minutes 14:38 partychat4: [EMAIL PROTECTED] Marnie/Aidan (is there a way to find out who is actually here using party chat? :p) ["aidan"] / list works ["Martin"] list is as close as you get :) [EMAIL PROTECTED] it would appear that an M4 RC will be rooled imminently... [EMAIL PROTECTED] rolled [EMAIL PROTECTED] any chance of getting the aptch in QPID-1469 in there ["aidan"] but that's not proper presence 14:39 [EMAIL PROTECTED] its 1 if statement, but it will stop people bringing the broker down on restart :P ["rhs"] what's QPID-1469? [EMAIL PROTECTED] if they use the management console ithout authentication ["aidan"] rhs: it's an if foo == null then don't NPE ["rhs"] sounds fine 14:40 ["rhs"] lemmie take a quick look... [EMAIL PROTECTED] oh, you are still here martin :) 14:41 ["rhs"] I don't have an issue with it. [EMAIL PROTECTED] the dentist didn't kill Martin today :-) 14:42 [EMAIL PROTECTED] lol :P 14:43 [EMAIL PROTECTED] Hi Rafi - any thoughts on a list of stuff to get to M4 finished state ? 14:47 ["rhs"] I don't believe there are any blockers at the moment. ["rhs"] gordon just killed the last one 14:48 ["Martin"] I'm always here Robbie, the log goes to my gmail account so I can check it even if i'm 'not here' :) 14:49 ["aidan"] hurrah! [EMAIL PROTECTED] ;) [EMAIL PROTECTED] ah, that's like synchronous messaging with the java broker .... 14:50 ["aidan"] robbie: is there a test for 1469? 14:51 [EMAIL PROTECTED] I believe that would be a no :) 14:52 ["aidan"] could there be a test? 14:53 [EMAIL PROTECTED] I just tested it manually. I havent actually looked at the tests yet to see what they do. [EMAIL PROTECTED] Rafi: Can you & Rajith update your M4 JIRAs pls ? [EMAIL PROTECTED] i think you're currently holding a combined total in the 50s :-) 14:56 ["aidan"] robbie: i think you could construct something around deleteUser() for it. ["aidan"] robbie: there doesn't seem to be anything existing [EMAIL PROTECTED] make a temp file, delete user, compare resulting file ? ["aidan"] robbie: yeah. the delete will likely chuck an exception without your patch 14:57 [EMAIL PROTECTED] indeed it will 15:00 ["aidan"] if that works out I'll commit it for M4, I've got something else I need to get in as well ________________________________ 6 minutes 15:07 partychat4: [EMAIL PROTECTED] should i just do it as a unit test on AMQUserManagementMBean, or against a live broker using jmx? ["aidan"] unit test please [EMAIL PROTECTED] k 15:08 ["aidan"] unless that turns out to be horrendous to set up, in which case you might find it easier to write a systest that uses QpidTestCase. A unit test should be simpler, and is preferable. [EMAIL PROTECTED] it is simpler, it just doenst necessarily test the method in question fully is all 15:09 [EMAIL PROTECTED] it will check the NPE doenst happen anymore though ["aidan"] that's the main thing, and that deleteUser works ["aidan"] though the second one is just a bit of a bonus ;) 15:10 [EMAIL PROTECTED] ;) 15:11 ["Martin"] I'd give bonus points to a systest aswell : 15:12 [EMAIL PROTECTED] how does QTC handle configs etc though ? since jmx is currently off by default :P 15:13 ["Martin"] Ah I have a JIRA open for that ) ["Martin"] Guess the Systest will need to wait for that. ["aidan"] martin: can i get you to review this when the commits gone in? ["Martin"] I'd like QTC.startBroker() to take a file which is the config ... but that really requires both C++ and Java brokers to be capable of using the same log file. ["Martin"] s/log/configuration/ 15:14 ["rob"] YAML! ;-) ["Martin"] Well having the ability for the test to say have use this acl file. ["aidan"] I want to kill our config file in the face, hard. ["Martin"] would be mighly useful. 15:15 ["Martin"] the main config file I'm happy to burn :)# ["rob"] I completely agree that the config files should have much more in common ["rob"] I just remeber the last time we had that conversation ["rob"] Also - our (Java) config files are hideous ["Martin"] When I finally get round to gettting some acl tests done then I'll be able to commit the new java acls that match the c++ so would be great to supply the acl via QTC to the broker for testing 15:16 ["Martin"] I don't think that the argument of 'we don't need config files' really works ["rob"] no ["aidan"] huh? ["rob"] Ah - probably before your time Aidan ["rob"] look for Yaml in the archives of qpid-dev 15:17 ["aidan"] no, thanks. ["rob"] :-) ["aidan"] i'll pass ["rob"] Also - certain people don;t believe in config files [EMAIL PROTECTED] ["Martin"] I don't think that the argument of 'we don't need config files' really works --- aconway makes this, it is not that you don't hav ethem, just that you THINK really hard before adding yet another option 15:18 ["Martin"] Aidan: If your doing the commit then sure I can review. Think it is important that we still follow the review process even for submitted patches. [EMAIL PROTECTED] so on the C++ broker, the config is the sum of the command line options / visa versa. ["aidan"] that makes sense ["Martin"] Carl so it would be good to standardise the options between brokers 15:19 ["rob"] Carl: I agree but only if it's not the case that you can configure things through opaque means (e.g. mgmt commands) that you can't configure in transparent config files ["rob"] the thing is more of a configuration database than a file anyway ["Martin"] true 15:20 [EMAIL PROTECTED] ack, but the point you make suports alan's position, stuff like mgnt commands should not be in the config file. ["aidan"] there are levers with knobs with dials ["rob"] There should be a way of viewing config that mgmt commands affects without starting the broker ["Martin"] I prefer toggles ;) 15:21 ["rob"] i.e. it needs to be in some readily readable form ["rob"] or tools provided to extract it [EMAIL PROTECTED] ["rob"] There should be a way of viewing config that mgmt commands affects without starting the broker -- that sounds like a mgnt client tool. 15:22 ["rob"] if the tool works directly on the DB of config information without the broker being up... then yes :-) [EMAIL PROTECTED] yes, or it may also be able to support a script / profiles. ["rob"] yep ["rob"] I think we are on the same page :-) 15:23 ["aidan"] the same confluence page? ;) 15:24 ["rob"] now that would be an idea :-) [EMAIL PROTECTED] yes - we are - there is an interesting project we use on our grid stuff called puppet.... it allow you to create config and then assert it or read it back, it manages the config in a repo ["Martin"] I someone going to wiki this so we can point to the page rather than write it again :) 15:25 [EMAIL PROTECTED] lol :) ["rob"] I think we need to make a conscious decision to dedicate time to harmonizing our approaches to config ... ["aidan"] yeah. we can only really do this sort of thing once 15:26 ["rob"] From the Java Broker point of view I think the first stage is to properly separate broker process config from model (queue/exchange/acl etc) config ["aidan"] so we need to get right or end up with sendmail ["rob"] :-) [EMAIL PROTECTED] rob, ack that is key, config should have nothing to do with the model 15:27 ["rob"] well model is a different sort of config... but they need to be kept distinct [EMAIL PROTECTED] ack. model config is valid, but not in the pid config file ["rob"] yep 15:28 [EMAIL PROTECTED] someone should cut this IRC and post it to list 15:29 ["Martin"] I was thinking more a wiki page that we could then use to workout what our configuration options would be.. much like we did for ACls ["rob"] if someone was really smart they'd write an agent that joined the party chat and e-mailed the list every 24 hours ... :-) 15:30 [EMAIL PROTECTED] if you want, we can start with the C++ config, and see what is relevant.... we have not model config in the C++ config files ["Martin"] That would be great. 15:31 [EMAIL PROTECTED] we can post the man page :-) on the wiki... ["Martin"] If the config options are not atleast documented on the wiki then that is a good starting point ["Martin"] :) ["aidan"] rob: odd you should mention that, I couldn't find anything when i googled by there are decent python xmpp bindings... ;) ["rob"] ;-) 15:33 ["Martin"] I could do that with some ease as I have somehow managed to get party chat to email my gmail account with the log ["Martin"] perhaps because I'm notionally logged in from my gmail account 15:34 ["aidan"] automated process would be better i think ["Martin"] well I'd just edit my gmail fitler to forward the partychat emails to qpid-dev :) ["aidan"] aaaah, cunning [EMAIL PROTECTED] if you have a compiled C++ broker, just do qpidd --help to get the configoptions ["aidan"] that'd be awesome 15:35 ["rob"] Martin: I wonder how often it would do that ["rob"] one mail per line would be annoying :-) [EMAIL PROTECTED] then each model loaded add's it's ownconfig options ["Martin"] I get around 2 emails a day ["rob"] yes - not quite sure how filters work against it though 15:36 ["rob"] as they obviously model it differently to real e-mails ["rob"] since the content keeps changing dynamically ["Martin"] I'll not apply my filter to the 500 chat msgs already in my in box but we'll see what the happens. ["Martin"] I've not received an email to day with this log ["rob"] ah - you already have a filter ["rob"] interesting 15:37 ["rob"] mine is all in the "chats" view ["Martin"] I probably have that too but I get real emails from the bot server -- Martin Ritchie
