--- Day changed Tue Apr 22 2008
00:00 < marainein> and of connexions
00:01 < kanzure> yes
00:01 < kanzure> http://heybryan.org/mailing_lists.html <-- added to the bottom.
00:02 < marainein> you don't find all that stuff hard to keep up with?
00:03 < kanzure> nope
00:09 < Kismet> i must say, some pretty intriguing and astonishing stuff collected on the wiki
00:10 < kanzure> Hi Kismet. :)
00:10 < kanzure> Didn't notice you coming in. Thought you were krebs.
00:11 < Kismet> hello, yeah i kinda came along from ##neuroscience
00:12 < kanzure> Yeah, have a look around the wiki, feel free to drop some comments or ask in here about wtf is going on :)
00:18 < Kismet> yeah i'm just reading and thinking "holy crap, this isn't as wildly impossible as i thought..."
00:20 < kanzure> Since you come from ##neuroscience, maybe you would be interested in the Makram computational neuroscience simulations? http://heybryan.org/intense_world_syndrome.html re: microcolumns and so on.
00:23 < Kismet> yikes, if only i didn't have to be to work first thing tomorrow
00:24 < Kismet> nighty night
00:24 -!- Kismet [n=chatzill@c-75-73-46-41.hsd1.mn.comcast.net] has quit ["ChatZilla 0.9.81 [Firefox 2.0.0.13/2008031114]"]
01:22 -!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has quit ["Leaving."]
02:42 -!- marainein_ [n=marainei@220-253-110-236.VIC.netspace.net.au] has joined #hplusroadmap
02:43 -!- marainein_ [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit [Client Quit]
03:08 -!- Aulere [n=dragon_d@131.229.176.252] has quit [Connection timed out]
03:29 -!- Aulere [n=dragon_d@131.229.176.252] has joined #hplusroadmap
03:47 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit ["Ex-Chat"]
05:47 -!- Aulere [n=dragon_d@131.229.176.252] has quit [Read error: 110 (Connection timed out)]
05:51 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has joined #hplusroadmap
07:19 -!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has joined #hplusroadmap
07:22 -!- Splicer [n=p@h6n1c1o261.bredband.skanova.com] has quit [Read error: 110 (Connection timed out)]
07:39 -!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has quit ["Leaving."]
07:52 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit ["Ex-Chat"]
08:44 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has joined #hplusroadmap
10:33 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit [Remote closed the connection]
17:08 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has joined #hplusroadmap
17:46 -!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has joined #hplusroadmap
19:23 -!- marainein [n=marainei@220-253-110-236.VIC.netspace.net.au] has quit ["Ex-Chat"]
19:57 -!- krebs_ [n=krebs@208.106.116.196] has joined #hplusroadmap
19:57 -!- krebs [n=krebs@208.106.116.196] has quit [Read error: 104 (Connection reset by peer)]
20:09 < kanzure> Hm.
22:16 < kanzure> fenn: Having git not work with debian etch, and relying on git for the repos, isn't a good idea since apt-get is supposed to *illustrate* its power ;)
22:16 < kanzure> soo I'm thinking we'll implement some repo interconversion tools
22:17 < kanzure> http://mwolson.org/projects/GitCvsSync.html git via a cvs sync
22:17 < kanzure> Anyway, I'm thinking of adding some ui hooks or something to agx-get, there'll be a dir where you drop in plugins and extensions, and basically this has to be heavily protected by the root super user
22:17 < kanzure> and any script added in there will be called at run time :-/
22:18 < kanzure> ideally these are 'bootup' scripts
22:18 < kanzure> such that they add in hooks and handlers for certain events
22:18 < kanzure> do you think this is a good idea/
22:18 < fenn> bootup scripts?
22:19 < kanzure> well, I mean that they are ran and then unloaded - they load up code into handlers or hooks, instead of running persistently with their own main script or something
22:19 < kanzure> dunno. I have to work out the architecture for this ...
22:19 < kanzure> basically the ikiwiki+git=>skdb thing is solid, except for review processes and authorization to upload a package - but basically it's done
22:20 < fenn> well, sorta
22:20 < fenn> we need some way to present a yaml file to the user for editing.. ikiwiki doesnt want to do anything but .mdwn files
22:23 < kanzure> http://developers.slashdot.org/article.pl?sid=07/06/03/004214 <-- Slashdot doesn't like git, apparently. They are suggesting svk.
22:23 < kanzure> hrm
22:23 < kanzure> why can't ikiwiki just know that everything in a certain dir is for editing
22:23 < kanzure> and then load up more plugins for content parsing / display over the web
22:29 < fenn> there is probably a plugin, i just havent looked yet
22:31 < kanzure> I don't know how I want to run the whole 'drop some plugins into agx-get' thing though
22:31 < kanzure> I mean, when do we demand agx-get is completely updated?
22:32 < kanzure> would we have them download a new script? or download a plugin that is an install script that writes over the previous version?
22:32 < kanzure> and then for new plugins to parse new metadata files? how would the output be done, how would we let new plugins ask the users for certain things or whatever?
22:33 < kanzure> there'd have to be a number of enforced modes where one could be "do something default and act like everything is okay" versus other modes where "ask the user on every single item" or "ask the user for some major stuff, but don't go all out" - and we'll just police the submitted plugins?
22:33 < fenn> er.. what would plugins be asking the user for?
22:35 < fenn> you could specify a requirement for a plugin version number in the .skdb yaml/metadata file
22:38 < fenn> slashdot is so full of inane banter and false basis for argument its hard to find anything worth worrying about
22:44 < kanzure> hm?
22:44 < kanzure> the plugins would be used to interpret new metadata files
22:44 < kanzure> so when the metadata is of a new type, you go fetch the necessary plugin to make agx-get work
22:45 < kanzure> but you don't want these plugins to do the entire user interface, do you?
22:45 < fenn> yes, plugins are not doing the requesting though, the program launching plugins to parse statements is requesting new plugins
22:45 < kanzure> so then how would it work
22:45 < kanzure> this sounds pretty simple in theory
22:46 < kanzure> uh, git a copy of the metadata only 
22:46 < kanzure> then search through it for the thing that the user wants
22:46 < kanzure> then parse the metadata file and give the user options?
22:46 < kanzure> or if the user has already specified parameters, chug through it
22:46 < fenn> you're reading along, reading along, then there's something you dont recognize. it should be the name of a plugin (i guess)
22:46 < kanzure> well, I thought it was the top of a yaml file
22:47 < kanzure> !!yaml/something/new/here/oh/my/god/
22:47 < fenn> the yaml file writer doesnt know what the user has installed
22:47 < kanzure> and then it'd check your sources.list and go find a package in skdb that offers that functionality for agx-get, and if not it gets a new set of 'metadata files' and finds it again
22:47 < kanzure> sure
22:47 < fenn> so you'd have to list every plugin used in the file, is that ok>
22:47 < kanzure> but agx-get can try loading it, but then it will fail
22:47 < kanzure> hrm
22:47 < kanzure> well
22:47 < kanzure> here's what I thought
22:47 < kanzure> I thought that the file was yaml itself
22:47 < kanzure> so therefore it's a serialized data structure
22:47 < fenn> yes
22:48 < kanzure> meaning that there's really only one class that you truly want to have that data in
22:48 < fenn> but the data doesnt mean anything without some code that can use it
22:48 < kanzure> and that's the class provided *by the plugin*
22:48 < kanzure> right
22:48 < kanzure> so the !!plugin/name/statement that is mandetory in yaml
22:48 < kanzure> is what we use, no?
22:48 < kanzure> it's actually a class name
22:48 < kanzure> but for us, we can conveniently use that for some other code as well
22:49 < fenn> i was thinking just use the name of the plugin, then the name of one of its members.. etc
22:49 < kanzure> see? we can have an skdb package with /that/class/name/ in file main.py, and then some other scripts that would be important for agx-get or something
22:49 < fenn> plugin:
22:49 < fenn>   member:
22:49 < kanzure> okay, so /plugin/name/member14141/
22:49 < fenn>    foo bar baz
22:49 < kanzure> nope
22:49 < kanzure> well
22:49 < fenn> why not?
22:49 < kanzure> because the entire yaml file is a serialization of a single class, no?
22:49 < fenn> no
22:49 < kanzure> that changes things
22:50 < fenn> it's a data structure, the data structure can have different code objects in it ( i think)
22:50 < kanzure> so can we request all sub-instances within the file that are of type "needsThisTypeOfPlugin" ??
22:50  * fenn looks up yaml types
22:50 < kanzure> uh
22:50 < kanzure> 'you think'
22:50 < kanzure> that's something rather important
22:51 < fenn> heh
22:54 < fenn> see example 2.24: http://yaml.org/spec/cvs/current.html
22:54 < fenn> it looks like you have to use an exclamation point, but the file structure can be mixed classes
22:54 < fenn> i'm fine with that
22:55 < kanzure> well, wait
22:55 < kanzure> worst-case scenario, we use a specific class that makes for an object with a list of names of plugins
22:55 < kanzure> and then another file for the actual content
22:55 < kanzure> but if mixed classes are allowed, or if the file parsing can be held off until we are sure we have operated per those plugin dependencies, then we're good
22:56 < fenn> the use of a hash of hashes was confusing me i guess
22:56 < kanzure> hm?
22:56 < fenn> plugin:
22:56 < fenn>   member:
22:56 < fenn> would translate to plugin=>member=>'foo bar baz'
22:57 < kanzure> I don't think that's useful
22:57 < kanzure> how about 
22:57 < fenn> oh, not quite
22:57 < fenn> 'plugin'=>'member'=>'foo bar baz'
22:57 < fenn> bah
22:57 < kanzure> pluginList
22:57 < kanzure>     memberUnit
22:57 < kanzure>         id: 349104914 mainRepo: fennetic.net/autogenix/autogenix.git
22:57 < kanzure> heh, my \n command was translated to plaintext over here
22:57 < kanzure> oh well
22:57 < kanzure> wait
22:57 < kanzure> was translated to nothing
22:58 < kanzure> whatever
23:04 < kanzure> what would be the best way to do it
23:04 < kanzure> would that allow for updates of all sorts?
23:04 < kanzure> and if so, what would the code to a plugin look like?
23:04 < kanzure> obviously we want agx-get to automatically take that 'pluginList' and check it against a list of known plugins
23:05 < kanzure> and then add the name of a plugin to a confirmed list of approved ones (just for security, it's not all that safe, but it's better than randomly accepting anything dropped in a directory)
23:05 < kanzure> what I am worrying about is conflicts between plugins
23:05 < kanzure> for example, how are they going to be processed?
23:05 < kanzure> which one gets executed first?
23:05 < kanzure> alphabetically? that's rather arbitrary
23:07 < fenn> could add plugin and sha1 hash
23:07 < fenn> for security
23:08 < fenn> what do you mean which gets executed first? they get instantiated in the order they appear in the metadata file
23:13 < kanzure> uhh
23:13 < kanzure> yes, that should have been obvious to me
23:14 < kanzure> I wonder where I put the python files I was working with a few days ago
23:15 < kanzure> ah, it was through IRC, I remember now
23:19 < kanzure> http://heybryan.org/mediawiki/index.php/2008-04-22
23:19 < kanzure> (a code highlighting extension for mediawiki might be a good idea)
23:19 < kanzure> probably GeSHi or something
23:21 < kanzure> http://www.mediawiki.org/wiki/Extension:Syntax_Highlighting