ruote tmp/log_2011-06-29.html

2011-06-29 06:24:31 utc jmettraux I hate amqp, I have to use ampq 0.6.7 and it defines its own BlankSlate, in the global namespace, which wrecks Parslet which uses the blankslate gem
2011-06-29 06:24:53 utc jmettraux later versions of amqp fix this issue, but I cannot upgrade :(
2011-06-29 06:25:36 utc tosch_le any chances of monkey-patching?
2011-06-29 06:27:18 utc jmettraux good idea
2011-06-29 06:28:58 utc jmettraux ah, I can work around it by require 'blankslate' before 'amqp'
2011-06-29 06:29:30 utc jmettraux ing
2011-06-29 06:30:25 utc jmettraux thanks for the "te" help yesterday
2011-06-29 06:33:33 utc tosch_le never mind. it was kschiess who thought about diving into the code.
2011-06-29 06:33:55 utc tosch_le which was, as always, helpful.
2011-06-29 06:34:14 utc jmettraux let's thank him too then
2011-06-29 06:34:24 utc jmettraux kschiess: hello, thanks for the 'te' help yesterday
2011-06-29 06:34:37 utc kitplummer (glad to be using ActiveMQ instead of RabbitMQ)
2011-06-29 06:35:30 utc tosch_le ACTION is not exactly glad, but perfectly happy using none of them ;-)
2011-06-29 06:35:37 utc jmettraux kitplummer: hello, you speak stomp to it ?
2011-06-29 06:35:53 utc kitplummer :) yep.
2011-06-29 06:36:49 utc kitplummer works great with ruby…there's a few bugs with JRuby preventing the same goodness on the JVM.
2011-06-29 06:36:56 utc kschiess service
2011-06-29 06:41:20 utc tosch_le would you think it's an abuse to use ruote as a background job service for rails?
2011-06-29 06:41:57 utc jmettraux if there is no workflow need, then yes
2011-06-29 06:42:52 utc kitplummer i think resque is pretty well equipped for just backgrounding (read async) jobs.
2011-06-29 06:43:14 utc tosch_le they are mini-workflows. beanstalkd would be great, but querying the status of finished jobs is not possible and would have to be implemented by me.
2011-06-29 06:44:33 utc kschiess How about keeping the jobs as DB entries as well, and reporting back to that?
2011-06-29 06:44:52 utc jmettraux à la DelayedJob
2011-06-29 06:45:42 utc kschiess It could be argued that you have two needs: a) job fulfillment, b) job tracking for the user.
2011-06-29 06:45:44 utc tosch_le delayed_job was an option, too, but it waits too long before actually starting the work and/or queries the db too often
2011-06-29 06:46:04 utc jmettraux :( ruote queries lots too
2011-06-29 06:46:30 utc tosch_le but nice little files when using the fsstorage
2011-06-29 06:46:39 utc jmettraux lol, yes
2011-06-29 06:47:13 utc tosch_le and i'd have every there and nearly ready (ok, that would be true for delayed_job, too).
2011-06-29 06:47:35 utc jmettraux and you have us all for the support
2011-06-29 06:48:03 utc tosch_le resque is no option because of redis: not yet another storage service, please ;-)
2011-06-29 06:48:05 utc kitplummer you should definitely look at resque.
2011-06-29 06:48:15 utc kitplummer :P ok, well nevermind.
2011-06-29 06:48:26 utc kitplummer but understood.
2011-06-29 06:49:28 utc kschiess tosch_le: ruote is much of a storage service as well...
2011-06-29 06:49:29 utc tosch_le that won't become a usual web service, but a kind of embedded device. there is quite a lot of memory and cpu power, but not that quite lot
2011-06-29 06:49:51 utc tosch_le kschiess: nah, it uses them.
2011-06-29 06:50:14 utc tosch_le kschiess: and it's one i'm used to.
2011-06-29 06:50:26 utc kschiess ACTION grins
2011-06-29 06:55:42 utc jmettraux kschiess: I just had an episode of amqp 0.6.7 break Parslet
2011-06-29 06:55:44 utc tosch_le ruote simply has everything i need for a proper background job service: great tracking of jobs, very decent error handling and it responds fast enough .
2011-06-29 06:56:37 utc jmettraux solved it by requiring 'blankslate' before 'amqp'
2011-06-29 06:57:04 utc kschiess And what was that about?
2011-06-29 06:57:35 utc jmettraux amqp defines its own BlankSlate in the global namespace (unless previously defined)
2011-06-29 06:57:50 utc jmettraux and that blankslate is not parslet friendly
2011-06-29 06:58:03 utc jmettraux tosch_le: bring on the issue reports !
2011-06-29 06:58:16 utc tosch_le :-)
2011-06-29 06:58:28 utc kschiess Name collision. The more I learn about everything amqp related, the less I like it.
2011-06-29 06:59:10 utc jmettraux later versions of the amqp gem are better, but I'm forced to use 0.6.7
2011-06-29 07:08:32 utc tosch_le_ thinking about the fsstorage… what about an INotifyWorker? it could watch for changes in the msgs directory and process new messages immediately when there are any. a sleep loop would only be necessary for scheduled actions
2011-06-29 07:09:45 utc jmettraux +1
2011-06-29 07:09:57 utc jmettraux (reminded me of https://github.com/guard/guard)
2011-06-29 07:11:32 utc jmettraux don't forget our JRuby friends :-)
2011-06-29 07:14:58 utc jmettraux beer is calling me
2011-06-29 07:16:03 utc jmettraux tosch_le_: inotify/fsevent whatever, it would be grand
2011-06-29 07:16:19 utc tosch_le_ let's see what i can do about it.
2011-06-29 07:16:52 utc tosch_le_ jmettraux: have fun drinking your beer!
2011-06-29 07:17:22 utc jmettraux thanks, it's like 37 out there, the first one will be pure hydration
2011-06-29 14:05:29 utc _Taz_ Could someone tell me how to setup a "process level variable" ? Or at least where I can find information about what is it/how it works?
2011-06-29 22:20:38 utc jmettraux memo _Taz_: if there is nobody replying here, please use the mailing list : http://groups.google.com/group/openwferu-users you set a process level variable by prefixing its name with a single slash (set 'v:/nada' => 'surf')