[05:01] *** Lawouach_ has joined #kamaelia |
[05:28] *** salmon_ has joined #kamaelia |
[06:38] *** Uraeus has joined #kamaelia |
[08:43] < Lawouach> hi folks |
[09:25] *** MS- has joined #kamaelia |
[09:25] < MS-> greetings |
[11:34] *** vmlemon_1 has joined #kamaelia |
[11:37] < vmlemon_1> Hi |
[11:41] *** vmlemon_2 has joined #kamaelia |
[11:50] *** jae_ has joined #kamaelia |
[11:53] < Lawouach_> back |
[11:56] < Lawouach_> MS-: I have a feeling it was for the best we didn't do GSoC this year |
[11:57] < Lawouach_> we all look busy busy already :) |
[11:57] < MS-> I think so. I'm actually planning on spending the time I'd spend on GSOC normally on actually improving kamaelia instead - which strikes me as a good idea :) |
[11:58] < Lawouach_> it does indeed :-) |
[11:58] < Lawouach_> Also, it sounds, going through of the mentor list, that it slowly becomes a heavy machinery with lots of bureaucracy |
[12:02] < MS-> Quite possibly. |
[12:04] < Lawouach_> anyhow, what's on your plate in regards to K. then? |
[12:05] < MS-> Well, getting Steve's code merged was one thing, but I've also been experimenting with a new axon core |
[12:06] < MS-> For which I've created a scratch google project as a place to dump my code ideas without cluttering up kamaelia's SVN |
[12:06] < MS-> That's currently sitting here: |
[12:06] < MS-> http://code.google.com/p/kamaelian/source/browse/sketches/Axon/ |
[12:07] < MS-> The difference really though is that I'm building up checking performance constantly along the way adding functionality in |
[12:07] < MS-> primarily so that I know the overall cost |
[12:07] < Lawouach_> sounds good |
[12:07] < MS-> Also, simon wittber's benchmarks give me a baseline |
[12:08] < MS-> I'm also planning on taking Jesse's comments about the website to heart and going to try and make the website *clearer* |
[12:08] < MS-> Which is always hard, since it's always* clear to the writer |
[12:08] < Lawouach_> it's very hard indeed |
[12:09] < MS-> I'm on leave for the next 2 weeks as well, so the first week I'll not do a huge amount, but the second week I plan to do a fair amount of kamaelia hacking. |
[12:09] < Lawouach_> Problem is you have one single landing page for many different people from various background and expectations |
[12:09] < MS-> Currently I'm rebuilding threaded components |
[12:09] < MS-> Yes, that is a key problem |
[12:09] < MS-> Jesse's talk at pycon - which spent ~2 minutes on kamaelia was helpful for making it clear what he thought is the problem |
[12:10] < MS-> As well as some follow ups since then |
[12:10] < MS-> (The video is online - I watched it last night) |
[12:10] < Lawouach_> http://divmod.org/trac < -- I think this is the kind of landing page I'd expect eventually |
[12:12] < MS-> Thing is I think that has the same problem from Jesse's perspective |
[12:12] < MS-> Suppose you know nothing about twisted for example |
[12:12] < MS-> where do you start on that page? |
[12:13] < Lawouach_> true |
[12:13] < MS-> You can see there's lots of cool stuff, and you know its good because lots of people say so - but where do you start |
[12:13] < Lawouach_> you probably need something at a higher level |
[12:13] < MS-> Which was the criticism of kamaelia |
[12:13] < MS-> his point of "pulling out and showcasing Axon" was more of a side note |
[12:14] < MS-> I think the point is the site needs to follow through common use cases, and help with those. |
[12:15] < MS-> It's the same sort of thing that Kathy Sierra has talked about for years as well. |
[12:15] < Lawouach_> ain't easy though considering how much stuff Kamaelia does |
[12:16] < Lawouach_> perhaps focusing on Axon would be better |
[12:16] < MS-> Her starting point is to suggest treating learning complex things like computerr games |
[12:16] < MS-> Indeed |
[12:16] < MS-> ie the idea being at each point people learn to do more. |
[12:18] < Patrick`> yeah |
[12:18] < Patrick`> I have to admit, it's a bit obtuse at times (the website) |
[12:18] < Patrick`> I would be available for consult as a source of reliable uninformedness |
[12:19] < Lawouach_> :) |
[12:19] < MS-> Patrick`: That's really helpful, believe it or not |
[12:19] < Patrick`> I know |
[12:19] < Patrick`> I've got tangled in stuff like that before |
[12:20] < Patrick`> When I took a summer job in the civil service, I redesigned the induction pack |
[12:20] < Patrick`> my most employable skill was that I was a noob |
[12:20] < MS-> I've also come across this BTW: Dreyfus Model of Skill Acquisition - which looks like a really useful guide for trying to understand skill levels |
[12:20] < MS-> (since teaching is really about moving between skill levels) |
[12:21] < MS-> I came across that here: http://oreilly.com/catalog/9781934356050/ (Pragmatic Thinking and Learning) |
[12:24] < MS-> BTW, regarding this new core: fibra results: |
[12:24] < MS-> "import hackysack" "hackysack.runit(10000,1000,0)" |
[12:24] < MS-> 10 loops, best of 3: 521 msec per loop |
[12:24] < MS-> vs New core results: |
[12:24] < MS-> "Hackysacker.runit(10000,1000,0)" |
[12:24] < MS-> 10 loops, best of 3: 339 msec per loop |
[12:24] < Patrick`> *high five* |
[12:24] < MS-> :) |
[12:25] < Patrick`> although optimising python for performance is like teaching a ballet dancer to punch harder |
[12:25] < MS-> New core is also threadsafe regarding scheduler allocation - to the extent of using STM |
[12:26] < Patrick`> ok, I give up |
[12:26] < MS-> heh |
[12:26] < Patrick`> a cursory google tells me what stm stands fo |
[12:26] < Patrick`> but what's the ... feeling of it, the shape of it |
[12:26] < MS-> STM == version control for variables |
[12:26] < Patrick`> we need a better word in english |
[12:26] < Patrick`> ah |
[12:26] < Lawouach_> MS-: impressive |
[12:26] < Patrick`> ok, I'm just wrong |
[12:27] < MS-> Lawouach_: I'm currently working on adding threads back in at an early level |
[12:27] < Lawouach_> right |
[12:27] < MS-> I can actually get a *much* faster result than that using the scheduler btw |
[12:27] < MS-> But it's not using normal kamaelia approaches |
[12:30] < MS-> To put it into perspective though the figures for Kamaelia 0.6.0: |
[12:30] < MS-> "import hackysack" "hackysack.runit(10000,1000,0)" |
[12:30] < MS-> 10 loops, best of 3: 5.59 sec per loop |
[12:31] *** vmlemon_1 has joined #kamaelia |
[12:31] < MS-> So about a 16 fold speed up |
[12:35] < Lawouach_> yeha |
[12:42] *** vmlemon_3 has joined #kamaelia |
[12:53] *** vmlemon_4 has joined #kamaelia |
[13:00] *** vmlemon_2 has joined #kamaelia |
[14:06] *** Davbo has joined #kamaelia |
[14:08] < Davbo> afternoon all |
[14:13] < vmlemon_2> Hi Davbo |
[14:25] *** jae_ is now known as eikenberry |
[15:34] *** vmlemon_1 has joined #kamaelia |
[15:37] *** jae_ has joined #kamaelia |
[17:43] *** jae_ is now known as eikenberry |
[18:07] *** jae_ has joined #kamaelia |
[18:07] *** salmon_ has joined #kamaelia |
[18:56] *** jae_ is now known as eikenberry |
[19:54] *** jae_ has joined #kamaelia |
[19:56] *** jae_ is now known as eikenberry |