2010-10-13 08:41:43 utc |
jmettraux |
lbt, gonzalo: I'm preparing a release of ruote-amqp 2.1.11, anything that I should be aware, that I should fix before that ? |
2010-10-13 08:42:26 utc |
lbt |
I'll take a look |
2010-10-13 08:42:45 utc |
jmettraux |
hello (btw) |
2010-10-13 08:43:06 utc |
lbt |
(hi!) we just found a bug that causes a failure in the AMQP code after 66000 processes |
2010-10-13 08:43:09 utc |
gonzalo |
hi john, i think the only issue we found is with event-machine and jruby |
2010-10-13 08:43:22 utc |
jmettraux |
gonzalo: hello, ah thanks |
2010-10-13 08:43:34 utc |
gonzalo |
no prob |
2010-10-13 08:43:38 utc |
lbt |
I *think* it's a problem in MQ not closing channels... |
2010-10-13 08:43:40 utc |
jmettraux |
lbt: hello, is that in the AMQP gem itself ? |
2010-10-13 08:43:55 utc |
lbt |
suspect so... so it shouldn't mattee |
2010-10-13 08:43:57 utc |
lbt |
r |
2010-10-13 08:45:15 utc |
jmettraux |
ouch, their issue list is quite long http://github.com/tmm1/amqp/issues |
2010-10-13 08:45:26 utc |
lbt |
yes... and idle |
2010-10-13 08:46:43 utc |
jmettraux |
:-( |
2010-10-13 08:47:27 utc |
lbt |
and whilst I realise my MQ patch was actually silly... I never heard back from the maintainer |
2010-10-13 08:47:45 utc |
jmettraux |
too bad |
2010-10-13 08:48:03 utc |
jmettraux |
he's quite accessible via IRC though |
2010-10-13 08:48:09 utc |
jmettraux |
tmm1 in #eventmachine |
2010-10-13 08:48:42 utc |
lbt |
OK |
2010-10-13 08:49:25 utc |
jmettraux |
http://github.com/tmm1/amqp/network maybe you'll have to resort to your own gem |
2010-10-13 08:49:48 utc |
jmettraux |
if you have some documentation about your issue, I'm sure kennethkalmer would love to see it |
2010-10-13 08:50:46 utc |
jmettraux |
if it's in ruote / ruote-amqp... let's fix it ! |
2010-10-13 08:51:35 utc |
lbt |
I agree |
2010-10-13 08:53:48 utc |
lbt |
this is my only patch http://github.com/lbt/ruote-amqp/compare/ruote2.1...register |
2010-10-13 08:54:09 utc |
lbt |
that url includes the ...register |
2010-10-13 08:54:18 utc |
jmettraux |
ouch |
2010-10-13 08:54:32 utc |
lbt |
I know... |
2010-10-13 08:54:47 utc |
lbt |
I'm happy to carry it |
2010-10-13 08:55:22 utc |
jmettraux |
let me think about it |
2010-10-13 08:55:50 utc |
jmettraux |
let me re-think about it rather |
2010-10-13 08:56:09 utc |
jmettraux |
thanks for the link |
2010-10-13 08:56:15 utc |
lbt |
I kinda think we need a way for the remote participant to communicate with the engine.... |
2010-10-13 08:56:33 utc |
jmettraux |
via the receiver ? |
2010-10-13 08:56:45 utc |
lbt |
yes.... but it may be nice to release 2.1.11 without rushing it |
2010-10-13 08:57:24 utc |
jmettraux |
wait wait, could you please rephrase "I kinda think we need a way for the remote participant to communicate with the engine...." |
2010-10-13 08:57:30 utc |
jmettraux |
is it about the register command ? |
2010-10-13 08:57:37 utc |
lbt |
yes |
2010-10-13 08:57:43 utc |
jmettraux |
ok |
2010-10-13 08:58:44 utc |
jmettraux |
what about passing such control messages to another consumer ? |
2010-10-13 08:58:57 utc |
jmettraux |
dedicated to control tasks ? |
2010-10-13 08:59:11 utc |
lbt |
yes... that would be fine |
2010-10-13 08:59:18 utc |
jmettraux |
whereas the vanilla ruote receiver only cares about workitem and launchitem |
2010-10-13 09:00:00 utc |
lbt |
FYI.... AMQP protocol multiplexes control and data messages over a single channel |
2010-10-13 09:00:11 utc |
jmettraux |
the "workflow" for you is it "I have to launch this process, but first I need to register participants" ? |
2010-10-13 09:00:20 utc |
lbt |
yes |
2010-10-13 09:00:37 utc |
jmettraux |
multiplex : OK, I didn't know, I was thinking about ruote data and ruote control (application level) |
2010-10-13 09:00:45 utc |
jmettraux |
ok |
2010-10-13 09:01:06 utc |
jmettraux |
would it make sense if I added something to register participants at the beginning of a process definition ? |
2010-10-13 09:01:22 utc |
lbt |
hmmm |
2010-10-13 09:01:23 utc |
jmettraux |
just for that instance ? |
2010-10-13 09:01:50 utc |
jmettraux |
(thanks for your time btw) |
2010-10-13 09:01:58 utc |
lbt |
I start remote participants as daemons... they register as being 'available' for work |
2010-10-13 09:02:09 utc |
jmettraux |
ok |
2010-10-13 09:02:22 utc |
jmettraux |
other variant |
2010-10-13 09:02:28 utc |
lbt |
I will allow users to register their own... |
2010-10-13 09:02:31 utc |
jmettraux |
what about a participant that registers particpants |
2010-10-13 09:02:42 utc |
lbt |
That's not a bad idea |
2010-10-13 09:02:48 utc |
lbt |
as it allows for validation |
2010-10-13 09:03:26 utc |
lbt |
I think I did that at first but thought it was hacky... now I think I may like it |
2010-10-13 09:03:50 utc |
lbt |
especially since it could deal with things like the catchall |
2010-10-13 09:04:02 utc |
lbt |
and "is this user allowed to register" |
2010-10-13 09:04:22 utc |
lbt |
but a 'trivial' use-case/process is "just register it" |
2010-10-13 09:05:49 utc |
lbt |
bbiab |
2010-10-13 09:16:00 utc |
jmettraux |
lbt: I have to move to another place, I'll be back soon, I'd like to give a try at this ParticipantManager[Participant] Prime Minister ? ttyl |
2010-10-13 09:39:14 utc |
rebo |
hey can anyone help with RuoteKit here? using the middleware in rails suppresses errors traces in the browser |
2010-10-13 09:52:49 utc |
rebo |
ok i figured it out by setting RuoteKit::Application::set(:show_exceptions,true) in ruote_kit.rb |
2010-10-13 09:53:04 utc |
rebo |
but would be nice to see the rails trace rather than the sinatra trace |
2010-10-13 09:55:04 utc |
jmettraux |
rebo: welcome to #ruote |
2010-10-13 09:55:15 utc |
rebo |
hey jmettraux |
2010-10-13 10:07:56 utc |
jmettraux |
lbt: (answer anytime) I'm doing the notebook work on the pm thing, |
2010-10-13 10:08:06 utc |
jmettraux |
so in your case, participants will register themselves |
2010-10-13 10:08:32 utc |
jmettraux |
what about the case when such a participant dies ? |
2010-10-13 10:10:57 utc |
jmettraux |
I know the pm thing itself hasn't to care about it, but I was wondering |
2010-10-13 10:44:41 utc |
jmettraux |
erlang-ish, the agent that did the registration does the supervision, and unregisters if needed |
2010-10-13 13:28:28 utc |
jmettraux |
lbt: http://gist.github.com/624031 |
2010-10-13 13:42:24 utc |
tosch_le_ |
rebo: sorry for answering so late. could please open an issue at http://github.com/kennethkalmer/ruote-kit/issues ? |
2010-10-13 13:42:48 utc |
tosch_le_ |
have no time at the moment to see into that but will do later |
2010-10-13 13:42:49 utc |
rebo |
hey tosch_le, no problem. Sure ill do that |
2010-10-13 13:46:19 utc |
rebo |
done |
2010-10-13 13:46:33 utc |
tosch_le_ |
thanks! |
2010-10-13 13:47:33 utc |
jmettraux |
:-) tosch_le, you'll be on it ? |
2010-10-13 13:48:14 utc |
tosch_le_ |
i'll try my best |
2010-10-13 13:48:17 utc |
tosch_le_ |
;-) |
2010-10-13 13:48:22 utc |
jmettraux |
great |
2010-10-13 13:48:23 utc |
rebo |
btw Ruote is pretty cool, but definitely hard to figure out! |
2010-10-13 13:49:58 utc |
jmettraux |
rebo: thanks, if you spot points to improve in the docs, please help |
2010-10-13 13:51:14 utc |
rebo |
I think the docs are kind of okay, but your missing some newby tutorials |
2010-10-13 13:51:26 utc |
rebo |
maybe when i've got this rails app working ill write something up |
2010-10-13 13:51:34 utc |
jmettraux |
true |
2010-10-13 13:52:23 utc |
rebo |
http://github.com/threetee/ruote-rails-example |
2010-10-13 13:52:49 utc |
rebo |
that example is pretty good, in that it shows how to use ruote in real situation |
2010-10-13 13:53:34 utc |
jmettraux |
real situation == rails ? |
2010-10-13 13:53:35 utc |
rebo |
though its rails 2.3 i think |
2010-10-13 13:53:39 utc |
rebo |
no |
2010-10-13 13:53:54 utc |
rebo |
i mean from my perspective im using rails |
2010-10-13 13:54:02 utc |
jmettraux |
ok |
2010-10-13 13:54:10 utc |
rebo |
but i mean the stuff in the docs is fairly abstract |
2010-10-13 13:54:50 utc |
jmettraux |
yes, it assumes people know what a workflow engine is |
2010-10-13 13:54:53 utc |
rebo |
yeah |
2010-10-13 13:55:08 utc |
jmettraux |
else they wouldn't search for one in the first place |
2010-10-13 13:55:17 utc |
jmettraux |
but anyway people get "lost" |
2010-10-13 13:55:22 utc |
rebo |
i think a lot of people who could make use of ruote dont really know about workflow engines |
2010-10-13 13:55:32 utc |
jmettraux |
true |
2010-10-13 13:55:33 utc |
rebo |
i mean i didnt till 2 days ago |
2010-10-13 13:55:37 utc |
rebo |
:D |
2010-10-13 13:55:52 utc |
rebo |
and its definitely useful for me |
2010-10-13 13:55:57 utc |
jmettraux |
lol |
2010-10-13 13:56:02 utc |
jmettraux |
great |
2010-10-13 13:56:20 utc |
jmettraux |
OK, points taken |
2010-10-13 13:57:36 utc |
rebo |
i think a lot of people are using statemachines where they should be using workflows |
2010-10-13 13:58:20 utc |
jmettraux |
+1 |
2010-10-13 14:20:13 utc |
jmettraux |
I should try to replicate that tutorial but with ruote http://www.ibm.com/developerworks/opensource/library/j-bpm1/index.html?ca=drs- |
2010-10-13 14:25:03 utc |
jmettraux |
lbt: I'll continue implementing this registerp thing tomorrow (especially the grabbing participant info from a workitem instead of from the process definition) |
2010-10-13 14:25:21 utc |
jmettraux |
if you have any questions / remarks / suggestions, please fire them on the mailing list |