ruote log_2010-03-15

2010-03-15 06:14:32 utc irfn Hi all
2010-03-15 06:14:38 utc jmettraux hi !
2010-03-15 06:15:29 utc irfn I had a question on "Participant"
2010-03-15 06:16:44 utc jmettraux please
2010-03-15 06:16:46 utc irfn A Step in a work flow can be a "Participant" where a User interaction is required to take the workflow further
2010-03-15 06:16:57 utc jmettraux yes
2010-03-15 06:17:09 utc irfn kool
2010-03-15 06:17:18 utc jmettraux Ruote::StorageParticipant usually
2010-03-15 06:17:39 utc irfn What is then a step which should run automatically without user interaction
2010-03-15 06:17:51 utc jmettraux it's a step too
2010-03-15 06:18:04 utc irfn ok
2010-03-15 06:18:08 utc jmettraux it all depends on which participant handles the "step"
2010-03-15 06:18:17 utc irfn :D
2010-03-15 06:19:27 utc irfn such as
2010-03-15 06:19:33 utc jmettraux http://groups.google.com/group/openwferu-users/browse_thread/thread/2c012627bdb5b1e7 last part of the thread is maybe interesting for you
2010-03-15 06:19:49 utc irfn oh ill read that first
2010-03-15 06:22:30 utc jmettraux if you're using ruote-kit and its REST interface, it's the part about /workitems
2010-03-15 06:33:06 utc irfn if i have a configured participant :bravo {|wi| puts "hi"}
2010-03-15 06:33:55 utc irfn if i have a sequence { participant :bravo }
2010-03-15 06:34:31 utc irfn it is a valid step of the workflow. rt?
2010-03-15 06:35:05 utc jmettraux yes
2010-03-15 06:36:31 utc irfn i have a decent example working with ruote
2010-03-15 06:36:40 utc jmettraux well done :)
2010-03-15 06:37:04 utc irfn but i am unable to get it working with ruote-kit
2010-03-15 06:37:22 utc irfn i have a very simple single step that i mentioned
2010-03-15 06:37:32 utc irfn and the storage is transient
2010-03-15 06:37:56 utc jmettraux what are the symptoms ?
2010-03-15 06:38:13 utc irfn the step doesnt get executed
2010-03-15 06:38:19 utc irfn i even did a raise in the step
2010-03-15 06:38:39 utc irfn just so that i am sure that its not a logging issue
2010-03-15 06:39:51 utc jmettraux how do you launch the process ?
2010-03-15 06:39:59 utc jmettraux what does the process look like exactly ?
2010-03-15 06:40:02 utc jmettraux http://gist.github.com/
2010-03-15 06:40:10 utc irfn ill do that
2010-03-15 06:40:18 utc irfn i just requiring a file at the end of config.ru in the route-kit gem source
2010-03-15 06:40:34 utc irfn just to get a rack app up quickly
2010-03-15 06:41:01 utc jmettraux ok, could you please gist that as well
2010-03-15 06:41:22 utc irfn doing that rt now
2010-03-15 06:42:05 utc irfn http://gist.github.com/332586
2010-03-15 06:42:19 utc jmettraux excellent
2010-03-15 06:44:01 utc jmettraux ok
2010-03-15 06:44:02 utc irfn when i change the config.mode = :file_system
2010-03-15 06:44:09 utc jmettraux wait
2010-03-15 06:44:16 utc irfn i and have the default catchall
2010-03-15 06:44:31 utc jmettraux you are registering the catchall before registering alpha and bravo
2010-03-15 06:44:47 utc jmettraux all your workitems are going to the catchall participant
2010-03-15 06:44:56 utc irfn yeah i guessed as much
2010-03-15 06:45:06 utc irfn but it doesnt work with the catchall commented
2010-03-15 06:45:09 utc jmettraux great
2010-03-15 06:45:23 utc jmettraux "it doesn't work" means what ?
2010-03-15 06:45:39 utc irfn the particpant that i registered
2010-03-15 06:45:42 utc jmettraux what are the symptoms ?
2010-03-15 06:45:43 utc irfn doesnt get called
2010-03-15 06:45:55 utc irfn i dont see the print statement
2010-03-15 06:46:04 utc irfn or if i change that to a raise
2010-03-15 06:46:10 utc irfn nothing bombs
2010-03-15 06:46:49 utc jmettraux so if you comment out line 26 of your config.ru, you're 100% affirmative that the behaviour stays the same ?
2010-03-15 06:46:57 utc jmettraux ie no visible output ?
2010-03-15 06:48:26 utc irfn yeah
2010-03-15 06:48:33 utc jmettraux yes or no ?
2010-03-15 06:49:00 utc irfn yes .. it still doesn't puts anything
2010-03-15 06:49:07 utc irfn or does it raise
2010-03-15 06:49:41 utc jmettraux if yes, then add this line somewhere after the engine got initialized :
2010-03-15 06:49:52 utc jmettraux RuoteKit.engine.context.logger.noisy = true
2010-03-15 06:50:00 utc jmettraux and please retry
2010-03-15 06:50:07 utc irfn ok
2010-03-15 06:50:23 utc jmettraux then please gist the output
2010-03-15 06:50:39 utc irfn ok will do that
2010-03-15 06:53:55 utc irfn i have updated the same gist http://gist.github.com/332586
2010-03-15 06:54:05 utc irfn with changes and the log output
2010-03-15 06:54:52 utc jmettraux very good
2010-03-15 06:54:59 utc jmettraux could you please place a
2010-03-15 06:55:05 utc jmettraux puts "here"
2010-03-15 06:55:13 utc jmettraux at the end of load_workflows.rb
2010-03-15 06:55:19 utc jmettraux to see if it gets executed ?
2010-03-15 06:59:08 utc irfn it does
2010-03-15 06:59:13 utc irfn just tried it
2010-03-15 06:59:27 utc jmettraux fishy
2010-03-15 07:01:08 utc jmettraux what are the six webrequests displayed in your log output ?
2010-03-15 07:02:11 utc jmettraux ok, trying here
2010-03-15 07:03:36 utc irfn we were trying the sinatra web interface to view if there are any processes
2010-03-15 07:04:23 utc irfn i think that shows only the persisted ones which have been caught via catcha_all
2010-03-15 07:05:22 utc jmettraux ok
2010-03-15 07:05:35 utc jmettraux got the same symptoms here
2010-03-15 07:05:37 utc jmettraux working on it
2010-03-15 07:08:03 utc irfn oh kool
2010-03-15 07:16:38 utc jmettraux that's weird, the worker doesn't seem to start
2010-03-15 07:16:54 utc jmettraux I have a similar application to yours and it works fine
2010-03-15 07:17:54 utc irfn :D
2010-03-15 07:18:09 utc irfn can you share that code
2010-03-15 07:18:18 utc jmettraux it's a full rails application
2010-03-15 07:18:40 utc irfn oh ok
2010-03-15 07:20:10 utc jmettraux that's what the initializer look like : http://gist.github.com/332599
2010-03-15 07:22:33 utc jmettraux OK, I'm going to study ruote-kit's code
2010-03-15 07:24:42 utc irfn :D
2010-03-15 07:25:09 utc jmettraux can't see anything fishy
2010-03-15 07:27:46 utc irfn here is a non ruote-kit equivalent code
2010-03-15 07:27:48 utc irfn http://gist.github.com/332602
2010-03-15 07:27:55 utc irfn without any storage
2010-03-15 07:28:00 utc irfn this works perfectly
2010-03-15 07:28:12 utc jmettraux good
2010-03-15 07:28:19 utc irfn i am tempted to not use ruote-kit
2010-03-15 07:28:19 utc jmettraux as should b
2010-03-15 07:28:20 utc jmettraux e
2010-03-15 07:28:43 utc jmettraux as you wish
2010-03-15 07:28:48 utc irfn but its gonna be a lot of work :D
2010-03-15 07:28:56 utc jmettraux I'm trying to fix / get that fixed anyway
2010-03-15 07:29:00 utc jmettraux thanks for raising the issue
2010-03-15 07:29:13 utc irfn no probs
2010-03-15 07:29:20 utc irfn and thanks for supporting us
2010-03-15 07:29:31 utc jmettraux you're welcome
2010-03-15 07:29:46 utc irfn i have a great deal of confidence with this now
2010-03-15 07:29:57 utc irfn we are surely gonna use ruote
2010-03-15 07:30:05 utc jmettraux cool
2010-03-15 07:36:57 utc jmettraux now it seems my rails application is running by pure luck
2010-03-15 07:42:02 utc jmettraux kennethkalmer: hi ! issue reports for ruote-kit, should they go to your copy or to Tosch's fork ?
2010-03-15 07:42:43 utc kennethkalmer jmettraux: hi !
2010-03-15 07:42:50 utc kennethkalmer erm, send them to mine
2010-03-15 07:42:56 utc jmettraux ok, thanks
2010-03-15 07:43:02 utc kennethkalmer tosch will also get notified :)
2010-03-15 07:47:05 utc jmettraux kennethkalmer, irfn: http://github.com/kennethkalmer/ruote-kit/issues/issue/4
2010-03-15 08:06:45 utc jmettraux tosch_le: hi !
2010-03-15 08:06:52 utc tosch_le hi!
2010-03-15 08:07:07 utc jmettraux sorry, but http://github.com/kennethkalmer/ruote-kit/issues/issue/4
2010-03-15 08:07:51 utc tosch_le just read the irc log
2010-03-15 08:08:07 utc tosch_le will have a look into it as soon as possible
2010-03-15 08:08:23 utc jmettraux ok, I couldn't locate anything
2010-03-15 08:08:41 utc jmettraux works like a charm in rails3
2010-03-15 08:08:59 utc jmettraux but rackup produces a no-worker engine
2010-03-15 08:23:36 utc tosch_le strange. no problems here
2010-03-15 08:23:45 utc tosch_le (vanilla rk)#
2010-03-15 08:24:51 utc jmettraux gem or github ?
2010-03-15 08:27:39 utc tosch_le github
2010-03-15 08:27:45 utc jmettraux me too
2010-03-15 08:35:02 utc jmettraux tosch_le : I placed that http://gist.github.com/332629 in an empty dir and ran bundle install
2010-03-15 08:35:05 utc jmettraux then rackup
2010-03-15 08:35:11 utc jmettraux but still, no worker :(
2010-03-15 08:35:36 utc jmettraux could it be because of ruote 2.1.8 ?
2010-03-15 08:35:45 utc jmettraux I guess you're running 2.1.7
2010-03-15 08:35:55 utc irfn yes
2010-03-15 08:36:03 utc irfn we are running 2.1.7
2010-03-15 08:37:21 utc jmettraux I'm running ruote 2.1.8 and I have the same issue...
2010-03-15 08:40:20 utc tosch_le same issue here using gem and ruote 2.1.8
2010-03-15 08:40:33 utc tosch_le so using the gem seems to be the culprit
2010-03-15 08:40:41 utc irfn oh
2010-03-15 08:40:59 utc irfn any way to get around this issue for now
2010-03-15 08:41:02 utc jmettraux I'm using github and ruote2.1.8 and I have the issue
2010-03-15 08:41:09 utc tosch_le oups.
2010-03-15 08:41:40 utc tosch_le a workaround would be to run the worker in a single process, but that's only a workaround, no solution
2010-03-15 08:42:20 utc jmettraux maybe the run_engine cleanup will fix that ;)
2010-03-15 08:42:48 utc irfn err..
2010-03-15 08:42:58 utc irfn sorry i didnt understand that
2010-03-15 08:43:05 utc tosch_le uncommented config.mode = :transient and there is a worker running
2010-03-15 08:43:14 utc irfn oh
2010-03-15 08:43:24 utc irfn because that defaults to file_system
2010-03-15 08:45:05 utc irfn so tosch_le
2010-03-15 08:45:15 utc irfn thats where we started off
2010-03-15 08:45:36 utc irfn we want a participant that needs no user intervntion
2010-03-15 08:45:50 utc jmettraux tosch_le: doesn't work for me
2010-03-15 08:45:57 utc jmettraux well strange
2010-03-15 08:45:58 utc tosch_le strange
2010-03-15 08:46:21 utc jmettraux there is activity, but the engine doesn't see the worker
2010-03-15 08:46:31 utc jmettraux engine.worker ==> nil
2010-03-15 08:46:46 utc irfn yeah i ran the code that you put on the gist
2010-03-15 08:46:50 utc irfn we can see the puts
2010-03-15 08:47:01 utc jmettraux ok
2010-03-15 08:47:06 utc jmettraux great
2010-03-15 08:48:23 utc jmettraux I guess I was mislead : I was expecting the engine to see the worker, but somehow ruote-kit starts a worker but doesn't connect the engine to it
2010-03-15 08:48:34 utc jmettraux so I was focusing on something wrong
2010-03-15 08:52:22 utc jmettraux yes, that's it
2010-03-15 08:52:29 utc tosch_le rk doesn't set the worker variable in ruote's engine
2010-03-15 08:52:38 utc tosch_le see run_worker!
2010-03-15 08:52:46 utc jmettraux ok
2010-03-15 08:53:14 utc jmettraux that means that the services are created two times
2010-03-15 08:53:19 utc tosch_le that's a no-go i suppose, as it breaks all instance level participants, i suppose
2010-03-15 08:53:28 utc jmettraux that some services
2010-03-15 08:53:53 utc tosch_le alas, where's the rewrite of run_engine! ? ;-)
2010-03-15 08:54:15 utc jmettraux lol
2010-03-15 08:56:37 utc jmettraux time for a refresh
2010-03-15 08:56:43 utc jmettraux the surface is good
2010-03-15 08:58:27 utc irfn here is a test workflow that i am looking at getting done via ruote and rk
2010-03-15 08:58:29 utc irfn http://gist.github.com/332638
2010-03-15 08:59:35 utc jmettraux rewind :if => '${f:http_status} == 422'
2010-03-15 09:00:19 utc jmettraux or
2010-03-15 09:00:43 utc jmettraux rewind :if -> '${f:http_status} != 200'
2010-03-15 09:00:56 utc jmettraux but that wouldn't work with 201 created
2010-03-15 09:11:38 utc irfn ok
2010-03-15 09:12:03 utc jmettraux irfn: you have a question about it ?
2010-03-15 09:12:16 utc irfn no
2010-03-15 09:12:47 utc irfn but i have been able to get a participant working that has no user intervation required.
2010-03-15 09:12:55 utc tosch_le i've rewritten RuoteKit.run_engine! a bit so that the worker is bound to the engine instance if configuration.run_worker is true
2010-03-15 09:13:05 utc irfn i mean i havent been able to
2010-03-15 09:13:06 utc tosch_le it's in my fork, could someone try that?
2010-03-15 09:13:12 utc jmettraux tosch_le: great !
2010-03-15 09:13:21 utc jmettraux irfn: great !
2010-03-15 09:13:48 utc irfn i ll give it a go
2010-03-15 09:14:04 utc tosch_le had no issues using the new code as gem, with transient storage and a block participant
2010-03-15 09:14:47 utc tosch_le (but didn't write a spec as i just didn't know how i could have specced that)
2010-03-15 09:15:25 utc jmettraux trying as well
2010-03-15 09:16:15 utc jmettraux gem 'ruote-kit', :git => 'git://github.com/tosch/ruote-kit.git'
2010-03-15 09:16:56 utc jmettraux looks fine
2010-03-15 09:19:56 utc jmettraux tosch_le : by the way, if you have time one of these days, I've severely updated ruote-fluo these days : http://github.com/jmettraux/ruote-fluo
2010-03-15 09:20:03 utc jmettraux if you want to update ;)
2010-03-15 09:20:54 utc irfn tosch_le: this fix works for us
2010-03-15 09:20:58 utc irfn tosch_le: thanks
2010-03-15 09:21:06 utc jmettraux irfn: many thanks !
2010-03-15 09:21:37 utc irfn tosch_le: jmettraux: many thanks guys !
2010-03-15 09:21:42 utc irfn much appreciated
2010-03-15 09:21:46 utc tosch_le thanks for reporting the bug. i'll release a new rk version, i suppose
2010-03-15 09:23:17 utc tosch_le john: thought of dropping RuoteKit.run! -- just don't see the use in it. i'll alias configure to it instead
2010-03-15 09:23:39 utc jmettraux tosch_le: OK
2010-03-15 09:25:31 utc jmettraux irfn: are you going to rubyconf india ?
2010-03-15 09:25:55 utc jmettraux you're a sponsor ;)
2010-03-15 09:28:55 utc irfn :D
2010-03-15 09:29:01 utc irfn i am gonna be there
2010-03-15 09:29:11 utc irfn i am not presenting anything though
2010-03-15 09:29:30 utc irfn lots of my colleagues are presenting
2010-03-15 09:29:31 utc jmettraux :(
2010-03-15 09:29:43 utc jmettraux there is an australian on this channel, biv who is going
2010-03-15 09:29:51 utc irfn why dont you come over
2010-03-15 09:30:05 utc irfn i am sure we can squeeze in a session
2010-03-15 09:30:07 utc irfn :)
2010-03-15 09:30:28 utc jmettraux ruote is boring ;)
2010-03-15 09:30:36 utc irfn lol
2010-03-15 09:30:41 utc jmettraux (with me as a speaker)
2010-03-15 09:30:48 utc jmettraux boring entreprisey
2010-03-15 09:31:25 utc irfn :)
2010-03-15 09:31:48 utc irfn well ruby/Rails has now become very enterprisey
2010-03-15 09:31:55 utc jmettraux +1
2010-03-15 09:31:59 utc irfn much to Matz distaste
2010-03-15 09:32:26 utc irfn poor guy! he wanted a nice fancy hobby language
2010-03-15 09:32:39 utc jmettraux lol
2010-03-15 09:33:01 utc irfn and dhh turns it into a platform of sorts
2010-03-15 09:33:03 utc irfn :)
2010-03-15 09:33:36 utc irfn atleast some excitement in my boring life. i mean compared to java
2010-03-15 09:33:39 utc jmettraux tosch_le: don't forget the latest ruote-fluo.js before releasing
2010-03-15 09:33:40 utc irfn :D
2010-03-15 09:34:10 utc tosch_le too late, sorry. but there's a time for rk 2.1.8.2 soon ;-)
2010-03-15 09:34:13 utc jmettraux irfn: you're in the right company. Is this project on jruby or c-ruby ?
2010-03-15 09:36:59 utc jmettraux have to change location, ttyl !
2010-03-15 10:36:19 utc tosch_le john: rk 2.1.8.2 is out including the latest ruote-fluo
2010-03-15 10:50:58 utc jmettraux back
2010-03-15 10:51:07 utc jmettraux tosch_le: woah, many thanks !
2010-03-15 10:51:17 utc tosch_le never mind.
2010-03-15 10:56:17 utc tosch_le i suppose i may close issue #4 now?
2010-03-15 10:58:32 utc jmettraux please do !
2010-03-15 11:02:16 utc tosch_le [x] done. thx for your investigation work
2010-03-15 11:02:53 utc jmettraux thanks to irfn, for me, I was just shooting myself in the foot
2010-03-15 11:18:31 utc kennethkalmer coffees ?
2010-03-15 11:18:39 utc jmettraux please
2010-03-15 11:19:25 utc kennethkalmer coming right up
2010-03-15 11:20:06 utc irfn hehe
2010-03-15 11:21:06 utc irfn i have interesting "Feature request" for ruote-kit
2010-03-15 11:21:13 utc jmettraux :)
2010-03-15 11:21:34 utc irfn perhaps ill do it and send a patch
2010-03-15 11:21:53 utc jmettraux fork ftw
2010-03-15 11:21:59 utc irfn so what do you guys think abt this,
2010-03-15 11:22:19 utc irfn rt now i can only see the running processes in the UI
2010-03-15 11:22:38 utc irfn what abt having an index of all the process definitions
2010-03-15 11:23:01 utc irfn and a POST link for each
2010-03-15 11:23:09 utc irfn to start a new process
2010-03-15 11:23:31 utc jmettraux that would mean that adding a definitions resource
2010-03-15 11:23:59 utc irfn sure
2010-03-15 11:24:25 utc jmettraux storage on disk ?
2010-03-15 11:25:43 utc jmettraux I'm not deciding for ruote-kit, but I was always rather cold for /definitions in ruote-kit
2010-03-15 11:26:18 utc jmettraux definitions could be stored in the company's DMS or in some web server, or created on the fly via a web application
2010-03-15 11:26:22 utc jmettraux so many possibilities
2010-03-15 11:27:08 utc kennethkalmer ACTION emits coffee
2010-03-15 11:27:16 utc jmettraux ACTION consumes
2010-03-15 11:28:25 utc kennethkalmer a few quick questions to get my rusty mind back inline
2010-03-15 11:29:39 utc kennethkalmer for rk catchall and couch I use the new StorageParticipant
2010-03-15 11:30:25 utc jmettraux me too
2010-03-15 11:31:15 utc kennethkalmer i'm actually lost, I can't believe it
2010-03-15 11:31:33 utc kennethkalmer i need to configure ruote-couch
2010-03-15 11:31:36 utc kennethkalmer for the engine
2010-03-15 11:31:46 utc kennethkalmer then use the StorageParticipant as the catchall
2010-03-15 11:32:17 utc jmettraux kennethkalmer: http://gist.github.com/332756
2010-03-15 11:32:55 utc jmettraux that's ruote[-couch] 2.1.8
2010-03-15 11:33:04 utc kennethkalmer thanks, just discovered set_storage()
2010-03-15 11:33:14 utc kennethkalmer it is amazing how quickly one can forget
2010-03-15 11:33:17 utc jmettraux kenneth, tosch_le: what do you think of irfn's proposal ?
2010-03-15 11:33:24 utc kennethkalmer well
2010-03-15 11:33:27 utc kennethkalmer i'm in too minds
2010-03-15 11:33:39 utc kennethkalmer i concur with you since i generate process definitions on the fly
2010-03-15 11:33:57 utc kennethkalmer however, it can be pretty useful to have /definitions load from disk
2010-03-15 11:34:14 utc jmettraux yes, simply put them in /public
2010-03-15 11:34:23 utc jmettraux problem solved
2010-03-15 11:35:01 utc kennethkalmer good point
2010-03-15 11:35:08 utc kennethkalmer i would make it read-only too
2010-03-15 11:35:17 utc kennethkalmer no management of the listing via ui
2010-03-15 11:35:26 utc jmettraux ACTION hopes /public is read-only
2010-03-15 11:35:27 utc kennethkalmer just read and launch
2010-03-15 11:35:30 utc kennethkalmer :)
2010-03-15 11:35:34 utc jmettraux :)
2010-03-15 11:35:51 utc jmettraux for my current app, I have /definitions
2010-03-15 11:35:52 utc irfn hmm
2010-03-15 11:36:01 utc irfn :)
2010-03-15 11:36:19 utc jmettraux http://gist.github.com/332764 (couch based)
2010-03-15 11:38:39 utc kennethkalmer neatly done
2010-03-15 11:40:00 utc jmettraux ruote-kit is perhaps a bit low-level to care about process definitions, it only cares about execution (IMHO)
2010-03-15 11:41:16 utc jmettraux /definitions is not in ruote-kit because end-users know better than us how they want it
2010-03-15 11:41:27 utc jmettraux that's debatable
2010-03-15 11:42:42 utc kennethkalmer actually, it can be added as middleware by the user
2010-03-15 11:42:48 utc kennethkalmer since we're in rack-world
2010-03-15 11:42:51 utc jmettraux +1
2010-03-15 11:43:13 utc kennethkalmer ok, so my final vote goes for making it a separate middleware
2010-03-15 11:43:18 utc kennethkalmer an example of what is possible
2010-03-15 11:43:27 utc kennethkalmer implementors can adapt it to their own liking
2010-03-15 11:43:30 utc kennethkalmer rk stays clean
2010-03-15 11:43:57 utc kennethkalmer s/clean/lean/
2010-03-15 11:45:48 utc irfn makes sense
2010-03-15 11:45:52 utc kennethkalmer irfn: apologies for my bad manners, how you doing ?
2010-03-15 11:46:06 utc irfn hi kennethkalmer
2010-03-15 11:46:13 utc irfn i am doing fine
2010-03-15 11:47:19 utc irfn thanks guys for your insights
2010-03-15 11:47:35 utc jmettraux you're welcome
2010-03-15 11:47:50 utc irfn i think i now understand rk enough to use it on my project
2010-03-15 11:48:52 utc jmettraux what kind of project is it ?
2010-03-15 11:49:26 utc irfn :)
2010-03-15 11:49:52 utc irfn so i have been working on this ruby prject for past 2 years
2010-03-15 11:50:10 utc irfn and we have built many apps for this customer
2010-03-15 11:50:19 utc irfn the apps follow a RADAR architecture
2010-03-15 11:50:48 utc irfn with a dump Webapp and a REST Engine (Service)
2010-03-15 11:51:24 utc jmettraux sounds cool
2010-03-15 11:52:24 utc irfn we already have scenarios where we call multiple Services to get work done
2010-03-15 11:52:29 utc irfn SOA style
2010-03-15 11:53:14 utc irfn we have code which is active resource calls to multiple engines
2010-03-15 11:53:24 utc irfn much of this is long running
2010-03-15 11:53:38 utc jmettraux :)
2010-03-15 11:53:39 utc irfn we have used RabbitMQ in a few cases
2010-03-15 11:54:00 utc irfn and now we are talking about this really big scenario
2010-03-15 11:54:18 utc irfn where we have some "orchestration" of services
2010-03-15 11:54:25 utc irfn which rollbacks etc
2010-03-15 11:54:38 utc irfn lots of error scenarios
2010-03-15 11:54:54 utc irfn so we thought abt using a workflow engines
2010-03-15 11:54:59 utc irfn *engine
2010-03-15 11:55:10 utc irfn and naturally ruote came in
2010-03-15 11:55:23 utc irfn we are currently working on a spike
2010-03-15 11:55:48 utc jmettraux in which sense ?
2010-03-15 11:55:57 utc jmettraux 'spike' ?
2010-03-15 11:56:16 utc irfn evaluate whether ruote can be used
2010-03-15 11:56:23 utc jmettraux ah ok
2010-03-15 11:56:53 utc jmettraux if you have any question, IRC + ML should get you covered
2010-03-15 11:56:54 utc kennethkalmer my 2c, if i may
2010-03-15 11:56:56 utc irfn we feel that ruote is good and can take care of our scenarios
2010-03-15 11:57:00 utc irfn sure
2010-03-15 11:57:22 utc irfn plz go ahead kennethkalmer:
2010-03-15 11:58:13 utc jmettraux even if you don't use ruote in the end, we'd be happy to hear any feedback, good or bad, we're eager to learn
2010-03-15 11:58:22 utc kennethkalmer i use ruote to orchestrate the provisioning of various desperate services across a wholesale ISP platform
2010-03-15 11:58:53 utc kennethkalmer the processes are scripted to embrace and handle failure
2010-03-15 11:59:07 utc kennethkalmer falling back to human participation where automation went wrong
2010-03-15 11:59:24 utc irfn that makes sense
2010-03-15 11:59:37 utc kennethkalmer and it works great
2010-03-15 11:59:48 utc kennethkalmer the day we went live, 3 guys in the office 'lost' their jobs
2010-03-15 11:59:54 utc kennethkalmer we had to figure out new things for them to do
2010-03-15 12:00:03 utc irfn whew
2010-03-15 12:00:10 utc irfn well
2010-03-15 12:00:21 utc kennethkalmer been like that since June 09, haven't looked back since
2010-03-15 12:00:49 utc irfn in our case client is gonna reduce a process which takes around 20 days
2010-03-15 12:00:56 utc irfn to around a few hours
2010-03-15 12:01:02 utc kennethkalmer awesome
2010-03-15 12:01:31 utc irfn this will allow them to bill their clients for those many days
2010-03-15 12:01:40 utc tosch_le ACTION is back from lunch (chinese)
2010-03-15 12:01:41 utc irfn yeah its awesome
2010-03-15 12:02:32 utc tosch_le on /definitions: would vote on not integrating that into rk, too. it depends too much on the use case and may be implemented by the user with low efforts
2010-03-15 12:03:50 utc irfn after the discussion here
2010-03-15 12:04:07 utc irfn i am of opinion that its best left to user to do this
2010-03-15 12:04:33 utc jmettraux :)
2010-03-15 12:07:42 utc tosch_le perhaps there should be some help to set remote_definition_allowed from within rk?
2010-03-15 12:08:12 utc jmettraux what about a well commented RuoteKit.configure {} example ?
2010-03-15 12:08:31 utc tosch_le that sounds perfectly ok
2010-03-15 12:08:34 utc jmettraux s/well/profusely/
2010-03-15 12:08:53 utc kennethkalmer :)
2010-03-15 12:09:15 utc irfn :)
2010-03-15 12:09:33 utc tosch_le volunteers to the front, please ;-)
2010-03-15 12:09:59 utc kennethkalmer right, so after weeks of distractions i finally have rk configured in iiab, onwards with the migration
2010-03-15 12:10:17 utc jmettraux congrats !
2010-03-15 12:10:25 utc kennethkalmer thanks :)
2010-03-15 12:10:31 utc kennethkalmer ah, wanted to ask
2010-03-15 12:10:35 utc kennethkalmer ruote-xmpp
2010-03-15 12:11:04 utc kennethkalmer you haven't touched it yet ? (or am I hallucinating)
2010-03-15 12:11:14 utc jmettraux I'm sorry, I never touched it
2010-03-15 12:11:28 utc jmettraux I have no use for it ;(
2010-03-15 12:11:39 utc kennethkalmer np, don't know where I got that thought from
2010-03-15 12:11:45 utc kennethkalmer i'll update it for 2.1
2010-03-15 12:11:56 utc kennethkalmer want to replace some of the amqp workers with xmpp workers
2010-03-15 12:11:56 utc jmettraux ruote-amqp is the host
2010-03-15 12:11:57 utc jmettraux hot
2010-03-15 12:12:05 utc jmettraux orly ?
2010-03-15 12:12:21 utc kennethkalmer well, the ones that leave the dc
2010-03-15 12:12:33 utc jmettraux wan ?
2010-03-15 12:12:40 utc kennethkalmer there is something brittle about amqp running through tons of firewalls
2010-03-15 12:12:44 utc kennethkalmer yeah
2010-03-15 12:12:59 utc kennethkalmer i have amqp workers on EC2, linode and in the UK
2010-03-15 12:13:19 utc jmettraux ouch
2010-03-15 12:13:47 utc kennethkalmer the ones running within Johannesburg are all fine
2010-03-15 12:13:58 utc kennethkalmer just when I leave the country.... :/
2010-03-15 12:14:52 utc jmettraux ruote-xmpp 2.1.x ftw !
2010-03-15 12:15:09 utc kennethkalmer based on blather, pure em-goodness :)
2010-03-15 12:15:20 utc jmettraux :)
2010-03-15 12:15:42 utc jmettraux btw : not sure about it, but http://github.com/jmettraux/ruote/blob/ruote2.1/lib/ruote/storage/composite_storage.rb#L36-41
2010-03-15 12:16:15 utc kennethkalmer which makes it possible to do messages over amqp/xmpp ?
2010-03-15 12:25:18 utc jmettraux ah no no
2010-03-15 12:25:42 utc jmettraux it just makes it possible to mix storages
2010-03-15 12:26:08 utc kennethkalmer was worth a shot :)
2010-03-15 12:26:37 utc jmettraux but yes, you could thus expressly manage msgs over ampq or xmpp
2010-03-15 12:26:44 utc jmettraux if there is a storage that does it
2010-03-15 12:29:42 utc kennethkalmer ACTION resists the urge to overcommit himself
2010-03-15 12:29:53 utc jmettraux don't do ti
2010-03-15 12:29:54 utc jmettraux it
2010-03-15 12:30:14 utc kennethkalmer i won't
2010-03-15 12:32:00 utc jmettraux kennethkalmer: question
2010-03-15 12:33:40 utc kennethkalmer shoot
2010-03-15 12:33:58 utc jmettraux ruote 2.1.x and passenger, OK ?
2010-03-15 12:35:10 utc kennethkalmer well, with or without a worker ?
2010-03-15 12:35:19 utc jmettraux with
2010-03-15 12:35:41 utc kennethkalmer not by default, but with some passenger tweaking it should work
2010-03-15 12:35:48 utc jmettraux :)
2010-03-15 12:37:17 utc kennethkalmer PassengerPoolIdleTime needs to be played with
2010-03-15 12:37:22 utc kennethkalmer there are more, just can't recall
2010-03-15 12:37:36 utc kennethkalmer trick is to ensure that there always at least one running copy
2010-03-15 12:46:26 utc irfn getting into meetings. bye thanks for all the help guys . ttyl.
2010-03-15 12:46:34 utc jmettraux ciao !
2010-03-15 13:40:08 utc kennethkalmer now I see the need for /errors
2010-03-15 13:41:33 utc jmettraux :)
2010-03-15 13:43:24 utc kennethkalmer http://gist.github.com/332847
2010-03-15 13:43:27 utc kennethkalmer any ideas ?
2010-03-15 13:44:33 utc kennethkalmer something to do with params
2010-03-15 13:46:43 utc jmettraux sorry was busy commenting : http://blog.codesherpas.com/on_the_path/2010/03/accounting-for-time-in-a-statebased-workflow.html
2010-03-15 13:48:13 utc jmettraux h.applied_workitem['fields'] holds a string it seems
2010-03-15 13:49:15 utc jmettraux not sure suddenly
2010-03-15 13:50:00 utc jmettraux ruby-1.9.1-p378 > 'toto'['nada'] = 'sruf'
2010-03-15 13:50:00 utc jmettraux IndexError: string not matched
2010-03-15 13:50:04 utc jmettraux it is
2010-03-15 13:50:46 utc jmettraux fields should be a {}
2010-03-15 13:50:51 utc kennethkalmer so it might be my launching that is faulty, will dig in further
2010-03-15 13:51:09 utc jmettraux what does your launch look like ?
2010-03-15 13:51:29 utc jmettraux the sig is Engine#launch(pdef, fields={}, variables={})
2010-03-15 13:51:42 utc jmettraux maybe you placed a string as the second param
2010-03-15 13:52:15 utc kennethkalmer yep, idiot...
2010-03-15 13:52:24 utc kennethkalmer {}.to_json instead of {}
2010-03-15 13:52:34 utc jmettraux aaah
2010-03-15 13:52:48 utc jmettraux remember Rufus::Json.encode and Rufus::Json.decode
2010-03-15 13:53:01 utc jmettraux will leverage yajl-ruby or json, whatever is present
2010-03-15 13:53:06 utc jmettraux yajl is the fast
2010-03-15 13:54:05 utc kennethkalmer have json
2010-03-15 13:54:13 utc kennethkalmer seems my ree segfaults with yajl
2010-03-15 13:54:33 utc kennethkalmer had to revert the yajl commit
2010-03-15 13:55:07 utc jmettraux ouch
2010-03-15 13:55:24 utc jmettraux Rufus::Json ftw
2010-03-15 13:56:43 utc kennethkalmer will have a look at what needs to be done to enable Rufus::Json as an ActiveSupport backend :)
2010-03-15 13:57:04 utc jmettraux nothing IIRC, it's already in
2010-03-15 13:57:05 utc kennethkalmer launch was successful after sending fields as a hash
2010-03-15 13:57:14 utc kennethkalmer ah, will give it a test
2010-03-15 13:57:35 utc jmettraux ActiveSupport::JSON is OK
2010-03-15 14:00:22 utc jmettraux http://github.com/jmettraux/rufus-json/blob/master/lib/rufus/json.rb#L40-43
2010-03-15 14:02:44 utc kennethkalmer i meant the reverse
2010-03-15 14:02:56 utc kennethkalmer allow AS::JSON to use Rufus::Json as a backend
2010-03-15 14:03:16 utc jmettraux ah, fun
2010-03-15 14:04:45 utc tosch_le about ruote on passenger: poolidletime has to be really large if there is low traffic
2010-03-15 14:04:54 utc jmettraux :)
2010-03-15 14:05:21 utc jmettraux I asked because of this state-based workflow blog post
2010-03-15 14:06:31 utc tosch_le and i'm unsure if there'll be trouble with passenger's smart spawning method: http://www.modrails.com/documentation/Users%20guide%20Apache.html#_smart_spawning_gotcha_2_the_need_to_revive_threads
2010-03-15 14:09:47 utc jmettraux from some rufus-scheduler discussion, I remember that the idle time tuning worked fine
2010-03-15 14:11:19 utc jmettraux maybe the world is split for ruote, if you need multiple front-end instances then you automatically qualify for (a) backend worker(s), if not, one process webserver is sufficient
2010-03-15 14:11:30 utc tosch_le i would run an own worker thread anyway, just to be on the safe side
2010-03-15 14:12:35 utc jmettraux I could write a cron-revived worker very quickly
2010-03-15 14:14:29 utc jmettraux we can have lots of fun
2010-03-15 14:15:25 utc kennethkalmer :)
2010-03-15 14:15:28 utc tosch_le :-)
2010-03-15 14:38:56 utc kennethkalmer jmettraux: workitem's put_at attribute, that is the same as the last_modified in 0.9 ?
2010-03-15 14:39:07 utc jmettraux yes
2010-03-15 14:39:15 utc kennethkalmer great
2010-03-15 14:39:30 utc kennethkalmer and the dispatch time, or is that not relevant anymore ?
2010-03-15 14:42:22 utc jmettraux can't remember at this one, let me grep
2010-03-15 14:42:32 utc jmettraux ACTION is with Amedeo on IM
2010-03-15 14:42:47 utc kennethkalmer brb - phone :/
2010-03-15 14:43:00 utc jmettraux dispatch_time seem to have vanished
2010-03-15 14:49:15 utc jmettraux please tell me if you need it, I guess we could add it as a field 'dispatch_time' or '_dispatch_time'
2010-03-15 14:53:11 utc kennethkalmer back
2010-03-15 14:53:26 utc kennethkalmer think it is good to have
2010-03-15 14:53:35 utc kennethkalmer to see how long it took to 'get here'
2010-03-15 14:53:56 utc jmettraux +1
2010-03-15 14:54:05 utc jmettraux __dispatch_time__ ?
2010-03-15 14:54:10 utc kennethkalmer +1
2010-03-15 15:01:11 utc jmettraux LogicallySound: hello Enrico
2010-03-15 15:01:16 utc LogicallySound Hey
2010-03-15 15:07:04 utc jmettraux LogicallySound: ruote 2.1.x is especially "un-brittle" when it comes to time based tests
2010-03-15 15:07:37 utc jmettraux kennethkalmer : http://github.com/jmettraux/ruote/commit/75c1d3aeea6aada1aff3117374f2305094bb3ee8
2010-03-15 15:07:45 utc LogicallySound jmettraux: Excellent. =)
2010-03-15 15:08:10 utc LogicallySound I'm looking forward to moving everything over to ruote 2.1
2010-03-15 15:08:16 utc LogicallySound Seems like it'll be much more robust and scalable.
2010-03-15 15:08:21 utc jmettraux :)
2010-03-15 15:08:27 utc jmettraux forking prawn ?
2010-03-15 15:08:46 utc LogicallySound Yeah, I was playing around with edge and I couldn't find a way of setting alignment of text in table cells.
2010-03-15 15:09:19 utc jmettraux I'll have to use it soon to generate reports, it seems a nice lib
2010-03-15 15:09:34 utc LogicallySound It's pretty slick.
2010-03-15 15:09:48 utc LogicallySound We're using it to generate invoices and pre-populated membership application forms.
2010-03-15 15:10:12 utc LogicallySound It isn't the best for large volumes of text that needs to be formatted and typeset, though it seems better than the alternatives there, too.
2010-03-15 15:11:16 utc jmettraux oh ok
2010-03-15 15:12:04 utc LogicallySound It has great table drawing features and the bounding box layout model is pretty solid.
2010-03-15 15:13:07 utc jmettraux I have to escape
2010-03-15 15:13:26 utc jmettraux kennethkalmer: welcoming your feedback on __dispatch_time__
2010-03-15 15:13:28 utc jmettraux ciao !
2010-03-15 15:20:06 utc kennethkalmer hey enrico !
2010-03-15 15:20:21 utc kennethkalmer i'll be back later, need to go do shopping and cook dinner