IRC Logs for #circuits-dev Monday, 2013-01-28

*** Osso has joined #circuits-dev01:11
prologichey Osso :)01:19
Ossohello :)01:19
prologicYou able to help fix the remaining 6 windows tests?01:19
prologicCheck jenkins' latest build01:20
prologicI fixed a few already01:20
prologicafter this we'll be fully tested on01:20
prologicLInux: py26, py27 py31, py3201:20
prologicWindows: py26, py27, py31, py3201:20
Ossowindows is not really my thing :(01:21
prologicneither is it mine :001:24
prologicyou don't have a windows box at all?01:25
prologiccause I don't :)01:25
prologicotherwise we'll have to leave it to Mark :)01:25
prologicwhich is fine01:25
prologicWhat I did want you to do however was read the conversion between ronny and myself on #circuits from the logs a couple days ago01:25
prologicand share your thoughts on the matter01:25
OssoI have desktop at home01:57
Ossowith windows01:58
Ossobut no dev tools installed01:58
Ossoand I have no experience in windows development01:58
Ossowho's ronny ?01:59
prologiche's a python dev that works with hpk a lot with pylib and pytest and other proejcts02:01
prologicfinished his phd recently02:01
prologicanyway read the discussions we had few nights ago02:01
prologicand tell me what you think :)02:01
OssoI am still not sure02:40
Ossowhat's missing02:41
Ossocomponent targeting was dealing with the speed issues02:43
Ossobut it never worked perfectly02:43
Ossowe can already start several objects graphs (one per thread)02:44
Ossoand we can already communicate between component in the same object graph using channels02:45
Ossoand I don't see how it is easier to have 2 object graphs in the same thread compared to have 2 channels02:46
prologicthis is what I wanted to discuss with you about03:39
prologicwhilst I sort of (very loosely) understand ronny's concerns03:40
prologiche's very much theoretical03:40
prologicand hasn't actually tried to use circuits to build what he has in mind03:40
prologicso this is what he wants to try to buidl on top of circuits03:42
prologicwhich actually is a concurrency model03:42
prologiccircuits doesn't implement the Actor model at al really03:42
prologicit implements pub/sub with async I/O multithreading and multiprocessing03:42
Ossoas I s see it03:45
Ossoit is kind of orthogonal03:46
Ossoyeah we don't implement actor model at all03:46
Ossobut you can send messages to other componants03:46
Ossowhich is you need to implement actors03:46
prologicyou wanna join in on this over at #circuits? :)03:46
prologicmehere, thanks :)03:55
prologicI feel I've said all I can03:55
prologicbut I think it's an interesting challenge03:55
prologicI might try to implement Actor in circuits03:55
prologicjust for fun of course03:55
prologicbe interesting to see if they have any value/benefit03:56
prologicpractical value/benefit that is03:56
prologicI've only ever seen Actor models in the wild a few times03:56
prologicand things implemented with an Actor based concurrency framework are very very slow03:56
prologicmehere, you are brilliant at explaining this :)03:59
prologicI must take a copy of the logs and massage that into a blog post!03:59
prologicSomething my late processor said that will always stick with me (where the ideas for circuits was drawn from)03:59
prologic"When you see an Airplane in the Sky, What do you think about?"04:00
prologic"I don't see an Airplane."04:00
prologic"I see a million components flying in coordination."04:00
mehereI like that. And I always try to believe in the coordination part when I have to fly.06:41
*** Osso has quit IRC07:45
*** Osso has joined #circuits-dev09:18
*** Osso has quit IRC09:29

Generated by 2.11.0 by Marius Gedminas - find it at!