[15:19:07] <dahoste> clearly, the task now is to figure out how to fake pypi downloads in order to game the pythonpackages.com leaderboard for fame and glory.
[15:19:27] <aclark> dahoste: the idea is if you are an author you can enter a package name
[15:19:40] <aclark> dahoste: if it exists, and it's downloads are hosted on pypi it will be "featured"
[18:18:26] <aclark> richleland: btw if we were ever to "take over" django district that doesn't necessarily mean we'd swallow it e.g. nova python. the goal is to facilitate and sustain as much activity as possible, not to genericize everything under the Python umbrella
[18:18:57] <aclark> i just jump on this stuff because i don't want to see zero activity
[18:19:22] <richleland> cool - absolutely understand - we just had some miscommunication last time so I figured I'd jump in front of that :)
[18:19:53] <richleland> totally agree with the zero activity, which is why I knew I needed to step aside
[18:20:02] <richleland> I just can't dedicate as much time as I used to
[18:20:23] <richleland> group deserves a more attentive leader
[18:22:11] <aclark> richleland: and our mission is to help anyone that organizes python activity. so e.g. if jackie doesn't want to pay to continue to host the meetup site we might be able to pay for it (i'm not sure how meetup billing works anymore… seems like we pay for one and organize three)
[18:22:21] <aclark> though pyladies has been covering some costs too which is awesome
[19:27:04] <aclark> it's great for deploying vain cheese to the clouds
[19:35:39] <hazmat> the price per process model seems a bit extreme to me.. is there a mem limit on a process?
[19:35:59] <hazmat> i recall its all lxc under the hood, so probably memset cgroup which allows bursting
[19:36:39] <hazmat> kennethreitz, i'm more curious in the context of comparison to cloud foundry, which basically offers the same notion of paas automation, if you deploy the paas
[19:37:38] <hazmat> kennethreitz, perhaps the better question... what does heroku offer to someone who's competent at setting up a linux box?
[19:38:23] <hazmat> a multi node cache/router? with hot failover db?
[19:42:12] <kennethreitz> sorry didn't see you there :0
[19:42:23] <kennethreitz> hazmat: correct, all soft limits, unless crazy
[19:42:39] <kennethreitz> i believe each dyno gets 512 mb or ram
[19:43:20] <kennethreitz> you setup your procfile, which is your list or processes
[19:43:32] <kennethreitz> and when you push, it gets deployed to a bunch of boxes
[19:43:39] <kennethreitz> and you tell it how many you want to run
[19:44:08] <kennethreitz> 'herou scale web=100', and you suddenly have 100, independent instances of your application, all running and load balanced automatically
[19:44:47] <kennethreitz> esentially you'd have to do a *lot* of working to do that yourself on top of ec2
[19:44:57] <kennethreitz> esp taking into consideration how faulty ec2 is
[19:48:45] <kennethreitz> well, good for 2007 maybe
[19:49:26] <hazmat> kennethreitz, not sure that really answers the compare to cloud foundry.. but ignoring that for a moment, how do you get app conn information for things like the pg db?
[19:49:30] <aclark> kennethreitz: so can i run pythonpackages.com (pyramid/redis) on heroku without changing any of teh codes?
[19:53:33] <kennethreitz> 'heroku config' just dumps out the configurations
[19:53:43] <hazmat> kennethreitz, when you say you mean you can use any db provider, you mean you can use anything you choose to manage yourself or get via a non heroku 3rd party
[19:53:56] <hazmat> else if you want the same management platform its postgres