ruote tmp/log_2011-05-06.html

2011-05-06 02:15:12 utc jmettraux lbt: hello, yes, rufus-json is a moody thing
2011-05-06 08:00:06 utc kennethkalmer coffees ?
2011-05-06 08:00:16 utc jmettraux oh please !
2011-05-06 08:00:34 utc kennethkalmer ACTION emits a beautiful organic espresso
2011-05-06 08:01:09 utc jmettraux ACTION consumes and says "many thanks !"
2011-05-06 08:01:18 utc kennethkalmer pleasure
2011-05-06 08:02:28 utc jmettraux I'm playing with http://kschiess.github.com/parslet/get-started.html it's excellent
2011-05-06 08:03:14 utc kennethkalmer wow, for the new dsl ?
2011-05-06 08:03:46 utc jmettraux yes, exactly
2011-05-06 08:04:25 utc kennethkalmer wow
2011-05-06 08:05:00 utc kennethkalmer i'm going to teach Platform45 (platform45.com) ruote in June for a whole week
2011-05-06 08:05:26 utc jmettraux oh wonderful
2011-05-06 08:05:40 utc kennethkalmer they have a project for a local bank's micro-lending division
2011-05-06 08:06:27 utc jmettraux I love their website
2011-05-06 08:06:42 utc kennethkalmer and IIAB will finally go live with a ruote 2.2 stack next week, I have roughly 20 more processes to run through
2011-05-06 08:06:48 utc kennethkalmer yeah, those guys are very good
2011-05-06 08:07:10 utc jmettraux I hope ruote 2.2 feels better than 1.9 for you
2011-05-06 08:07:18 utc kennethkalmer oh man
2011-05-06 08:07:21 utc kennethkalmer much better
2011-05-06 08:07:27 utc kennethkalmer lighter
2011-05-06 08:07:30 utc kennethkalmer easier to work with
2011-05-06 08:07:37 utc jmettraux w00t !
2011-05-06 08:07:45 utc kennethkalmer and having rk in the rails app makes a huge difference
2011-05-06 08:07:55 utc kennethkalmer local participants can directly interact with the models
2011-05-06 08:08:28 utc kennethkalmer the legacy code used a ton of rest calls, i've removed so many actions from our controllers, more than the code of the new participants
2011-05-06 08:08:36 utc kennethkalmer very very happy overall :)
2011-05-06 08:08:55 utc jmettraux sweet, let me know if there is anything we can improve
2011-05-06 08:09:30 utc kennethkalmer so far nothing :)
2011-05-06 08:09:39 utc kennethkalmer just this tiny patch for ruote-amqp
2011-05-06 08:10:21 utc kennethkalmer https://gist.github.com/958594
2011-05-06 08:10:22 utc jmettraux maybe you could commit and point to the commit if you use bundler
2011-05-06 08:10:51 utc kennethkalmer the commit doesn't break anything now, it is a drop-in replacement
2011-05-06 08:10:59 utc kennethkalmer let me fire it off to the list again
2011-05-06 08:12:45 utc jmettraux ACTION re-reading the thread
2011-05-06 08:14:11 utc kennethkalmer David got a bit carried away imho, stuff worth considering for later though
2011-05-06 08:15:59 utc jmettraux he's "lbt" here, wondering if he's awake
2011-05-06 08:16:10 utc lbt kennethkalmer: hey
2011-05-06 08:16:21 utc kennethkalmer hi hi hi lbt
2011-05-06 08:16:27 utc jmettraux hello lbt
2011-05-06 08:17:06 utc lbt so ... ruote-amqp changes :)
2011-05-06 08:17:47 utc kennethkalmer :)
2011-05-06 08:18:05 utc kennethkalmer first thing is just that tiny patch
2011-05-06 08:18:11 utc lbt url ?
2011-05-06 08:18:41 utc kennethkalmer https://gist.github.com/958594
2011-05-06 08:18:44 utc lbt I didn't want to have to change all the participants I've written and my "customers" have written
2011-05-06 08:18:59 utc kennethkalmer I fully agree
2011-05-06 08:19:05 utc lbt ok .... so move to consume()
2011-05-06 08:19:08 utc lbt not remove :)
2011-05-06 08:19:13 utc kennethkalmer yes
2011-05-06 08:19:17 utc kennethkalmer your vito was brilliant
2011-05-06 08:19:18 utc lbt phew
2011-05-06 08:19:26 utc kennethkalmer got me thinking again
2011-05-06 08:19:30 utc lbt I thought you meant move out to user-code
2011-05-06 08:19:37 utc lbt silly me
2011-05-06 08:19:53 utc kennethkalmer initially the proposal was to let the user call RuoteAMQP.start! on their own
2011-05-06 08:19:58 utc kennethkalmer but your right
2011-05-06 08:20:09 utc tosch_le hi all
2011-05-06 08:20:13 utc kennethkalmer why deploy new code when a drop-in replacement just works
2011-05-06 08:20:16 utc kennethkalmer hi tosch_le!
2011-05-06 08:20:22 utc lbt hi tosch_le
2011-05-06 08:20:58 utc tosch_le what i though when i read the initial mail i asked myself: why not call RuoteAMQP.start! only when there is a worker present?
2011-05-06 08:22:12 utc tosch_le but then i answered myself: registration is done only in one engine…
2011-05-06 08:22:34 utc kennethkalmer true
2011-05-06 08:22:35 utc jmettraux hello tosch_le
2011-05-06 08:23:06 utc lbt It also struck me as a rather advanced use-case ... and I wondered if it would be better 'explicit'
2011-05-06 08:24:24 utc kennethkalmer brb - just getting cheese for omelets :)
2011-05-06 08:24:32 utc lbt when does a p call consume() .. let me look
2011-05-06 08:28:09 utc lbt was wondering quietly if it matters that the amqp q is only created when the first wi is sent out
2011-05-06 08:30:11 utc kennethkalmer it does in some cases
2011-05-06 08:30:29 utc kennethkalmer in this particular case of mine, i'm using rk with passenger
2011-05-06 08:30:37 utc lbt I can see situations in failure/recovery where that could happen
2011-05-06 08:30:39 utc kennethkalmer then worker runs in a rake task
2011-05-06 08:30:43 utc lbt so if it matters I think it's a bug
2011-05-06 08:30:54 utc lbt (I meant in a bad way)
2011-05-06 08:31:21 utc lbt you need it to not happen on a certain machine/process :)
2011-05-06 08:31:41 utc kennethkalmer i don't want any amqp connections setup in the passenger instances
2011-05-06 08:31:41 utc kennethkalmer yep
2011-05-06 08:31:42 utc kennethkalmer the amqp connection is only required in consume() in any case
2011-05-06 08:31:49 utc kennethkalmer not in the initializer
2011-05-06 08:32:26 utc lbt of course there's the massive overhead of a call to RuoteAMQP.start! in every consume ;)
2011-05-06 08:32:56 utc lbt receiver?
2011-05-06 08:33:03 utc kennethkalmer haha :)
2011-05-06 08:33:12 utc kennethkalmer receiver can only function if an amqp connection exists
2011-05-06 08:33:26 utc lbt so no change there
2011-05-06 08:33:30 utc kennethkalmer nothing
2011-05-06 08:33:37 utc lbt cool - I'm happy
2011-05-06 08:33:45 utc kennethkalmer w00t !
2011-05-06 08:33:54 utc lbt I have this deployed in Nokia, Intel, and MeeGo.com FYI
2011-05-06 08:34:04 utc lbt and it's in heavy use
2011-05-06 08:34:15 utc kennethkalmer 10.times { w00t! }
2011-05-06 08:34:15 utc lbt so "thanks" .... really!
2011-05-06 08:34:36 utc kennethkalmer and thanks for your vote of confidence in the team's abilities
2011-05-06 08:34:39 utc jmettraux ACTION cheers
2011-05-06 08:34:47 utc lbt did I ever show you where I'd like ruote-amqp ?
2011-05-06 08:35:21 utc kennethkalmer do share
2011-05-06 08:35:27 utc lbt http://wiki.meego.com/Release_Infrastructure/BOSS/Design
2011-05-06 08:35:41 utc lbt old slides .. still like the idea
2011-05-06 08:35:52 utc lbt I want to enable the out-of-band messaging
2011-05-06 08:36:00 utc lbt for cancel/pause/resume
2011-05-06 08:36:19 utc lbt and I think creating a queue with the wfid as the queue name would work
2011-05-06 08:37:19 utc lbt a typical long-lived remote participant would have 2 threads - one busy doing stuff and another listening
2011-05-06 08:39:10 utc kennethkalmer when using wfid as the queue names, how do you "advertise" your available queues ?
2011-05-06 08:40:00 utc lbt you do an amqp "don't create q" and catch the exception ?
2011-05-06 08:41:25 utc lbt I didn't want to have to notify ruote and have it keep info on who listened where
2011-05-06 08:42:23 utc lbt this is on my todo list - but you may have heard that Nokia pulled out of MeeGo .... kinda screwed up my plans :)
2011-05-06 08:42:36 utc kennethkalmer sigh
2011-05-06 08:42:48 utc lbt yeah
2011-05-06 08:43:19 utc lbt in truth it may mean more smaller meego users - and that may be better for the whole ecosystem
2011-05-06 08:43:37 utc lbt we have china mobile, LG, samsung all using it
2011-05-06 08:43:47 utc kennethkalmer cool
2011-05-06 08:44:22 utc lbt and I'm presenting 3 sessions which are BOSS (and hence ruote/amqp) oriented in our meetup in SF
2011-05-06 08:44:47 utc lbt bear in mind about 140 sessions got turned away... 3 on this stuff is quite serious interest
2011-05-06 08:44:54 utc kennethkalmer awesome
2011-05-06 08:45:03 utc kennethkalmer let us know when the slides are up :)
2011-05-06 08:45:34 utc lbt I will - though the focus is on the business side of process driven automation
2011-05-06 08:46:06 utc kennethkalmer that is very important too
2011-05-06 08:46:20 utc kennethkalmer think that is the biggest selling point for ruote
2011-05-06 08:46:32 utc kennethkalmer as business friendly as it is developer friendly
2011-05-06 08:46:42 utc lbt *nod*
2011-05-06 08:47:02 utc lbt And the radial thing should help
2011-05-06 08:47:34 utc lbt I don't want to show it often ... but it's nice if it looks clean when we do
2011-05-06 08:47:51 utc jmettraux working on it
2011-05-06 08:47:59 utc kennethkalmer :)
2011-05-06 08:48:12 utc lbt well, I have 2.2.1~rc1 in meego.com now
2011-05-06 08:49:01 utc lbt kennethkalmer: hmm ... wondering about showing you skynet
2011-05-06 08:49:17 utc kennethkalmer the map/reduce gem ?
2011-05-06 08:49:17 utc lbt remote participant manager at the OS level (daemontools)
2011-05-06 08:49:23 utc jmettraux the skynet bash script you showed me yesterday was excellent
2011-05-06 08:49:25 utc kennethkalmer oh wow
2011-05-06 08:49:59 utc lbt http://wiki.meego.com/Release_Infrastructure/BOSS/Installationhave a look at:
2011-05-06 08:50:09 utc lbt err have a look at http://wiki.meego.com/Release_Infrastructure/BOSS/Installation
2011-05-06 08:50:27 utc lbt especially the "Getting started" section at the end
2011-05-06 08:51:42 utc lbt that's how users deploy and manage participants
2011-05-06 08:52:22 utc kennethkalmer epic
2011-05-06 08:52:46 utc lbt kennethkalmer: nice things: runs as any user (externalised suid). logging is a different user. code doesn't know it's a daemon. print to stdout to log
2011-05-06 08:53:06 utc kennethkalmer +1
2011-05-06 08:53:24 utc lbt if you deploy this single file : http://pastie.org/1866041 (or a ruby equiv)
2011-05-06 08:54:16 utc lbt I know you have the daemonise library so you may be interested in an alternative
2011-05-06 08:54:37 utc kennethkalmer that is awesomely simple
2011-05-06 08:54:54 utc lbt thanks
2011-05-06 08:55:03 utc kennethkalmer one last thing for ruote-amqp
2011-05-06 08:55:09 utc kennethkalmer the dependency specification for ruote
2011-05-06 08:55:10 utc lbt yep
2011-05-06 08:55:18 utc kennethkalmer can I make it -> s.add_runtime_dependency 'ruote', "~> 2.2.0"
2011-05-06 08:55:30 utc lbt I think that's fine
2011-05-06 08:55:43 utc lbt that's fine :)
2011-05-06 08:56:01 utc lbt the BOSS suite is a new setup and will likely use 2.2.1
2011-05-06 08:56:17 utc kennethkalmer jmettraux: wdyt ?
2011-05-06 08:57:42 utc jmettraux kennethkalmer: sounds good... Aren't you using bundler ?
2011-05-06 08:58:17 utc lbt eep
2011-05-06 08:58:28 utc kennethkalmer I am, wanna release 2.2.0.1 or should we skip it ?
2011-05-06 08:59:23 utc lbt ACTION notes that he also has to repackage all dependencies when ruote et al changes them. It would be nice to pick up dep changes that are needed.
2011-05-06 09:00:34 utc lbt eg iirc 2.2.1 changes a few versions in the underlying libraries .... not sure if they're really needed or just picked up as "latest"
2011-05-06 09:00:35 utc jmettraux sorry lbt, our rubygemers/bundlers life is so paradisiac
2011-05-06 09:01:36 utc lbt yeah ... it makes it hard to have other things running on the deployed box ... usual dev/sysadmin/security issues
2011-05-06 09:02:01 utc lbt not worth getting into ... and I don't know how hard my request is
2011-05-06 09:02:32 utc kennethkalmer ok, let me do this then, just push the one patch to github and leave it at that
2011-05-06 09:02:58 utc kennethkalmer we can do the version bump formalities with 2.3 ?
2011-05-06 09:03:06 utc kennethkalmer along with any other refactoring required (if any)
2011-05-06 09:04:06 utc lbt jmettraux: don't get me wrong ... I have no problem at all using newer deps .... but it would be nice if they were only newer if new features were needed.
2011-05-06 09:04:26 utc lbt Anyhow :) - I'll let you get on .... thanks for talking about the API (non-) change. kennethkalmer: If you get to thinking about the out-of-band stuff then I'd appreciate a chat :)
2011-05-06 09:04:52 utc kennethkalmer will let my mind dwindle on it for a few days :)
2011-05-06 09:05:07 utc kennethkalmer https://github.com/kennethkalmer/ruote-amqp/commit/f049ad962f4abd52700c6909351af118828bd9fe
2011-05-06 09:05:44 utc jmettraux lbt: do you package gem by gem ?
2011-05-06 09:05:56 utc lbt jmettraux: yes
2011-05-06 09:06:14 utc jmettraux what about using bundler to draw the whole lot in and then package the "bundle" ?
2011-05-06 09:06:15 utc lbt I'm working with the debian-ruby guys on their automation tool though
2011-05-06 09:06:48 utc lbt jmettraux: don't forget ... the main users are distro-maintainers... *g*
2011-05-06 09:06:56 utc jmettraux debian-ruby : do you have one japanese guy among those ?
2011-05-06 09:06:57 utc lbt they would actually take a hit out on me
2011-05-06 09:07:06 utc jmettraux ok
2011-05-06 09:07:23 utc lbt jmettraux: I don't know them that well at all yet
2011-05-06 09:07:31 utc jmettraux ok
2011-05-06 09:07:39 utc lbt some good stuff ... but still some issues
2011-05-06 09:07:50 utc lbt eg they insist on renaming gems :(
2011-05-06 09:46:35 utc kennethkalmer man, ruote can be a nosql store as well, just like solr :)
2011-05-06 09:53:16 utc kennethkalmer jmettraux: i'm picking up a pattern during refactoring of process defs
2011-05-06 09:53:40 utc kennethkalmer i had an old remote participant, particularly unreliable
2011-05-06 09:53:55 utc kennethkalmer everywhere i called it, i had a :timeout param specified
2011-05-06 09:54:15 utc kennethkalmer wonder if it could be useful during participant registration to specify a timeout
2011-05-06 09:55:23 utc kennethkalmer most of my autonomous remote participants have timeouts, and fall back to human for intervention in the event of a timeout
2011-05-06 14:22:52 utc kennethkalmer chsh -s /usr/local/bin/zsh
2011-05-06 14:23:00 utc kennethkalmer lets see how this goes over the coming days
2011-05-06 14:27:58 utc jmettraux good luck, tell me how rvm fares
2011-05-06 14:29:20 utc kennethkalmer so far perfactly
2011-05-06 14:29:24 utc kennethkalmer *perfectly
2011-05-06 15:06:25 utc jmettraux have a good evening !