Active2 Dev Meeting Log, 8 May 2002
Session Start: Wed May 08 11:57:48 2002
Session Ident: #active2
[11:57] *** Now talking in #active2
[11:57] *** Topic is 'active2 meeting at 1900 UTC -- http://docs.indymedia.org/view/Devel/MeetingAgenda2002May08'
[11:57] *** Set by firstname.lastname@example.org on Wed May 08 11:49:14
[11:58] <scott-van> hey keith... wondering where you were ;-)
[11:58] <vic_keith> was asleep
[11:58] <scott-van> mike had to step away for a sec
[11:58] <vic_keith> kay
[11:58] <vic_keith> <yawn>
[11:59] <scott-van> saw to started work at 10:15 last night... when did you crash?
[11:59] <vic_keith> 5:30 am
[11:59] <vic_keith> been pulling some late shifts...
[11:59] <scott-van> eww... an all-nighter
[12:00] <scott-van> hey are coming to CPSR next week?
[12:00] <vic_keith> spent a couple hours scanning a cookbook (remember book-piracy?)
[12:01] <vic_keith> re cpsr: yes.. i have to talk to yuo about that \
[12:01] <scott-van> my memories of book piracy center around trying to photocopy AD&D
adventures with the light blue (unphotocopiable, supposedly) maps ;-)
[12:01] <vic_keith> ha
[12:01] <vic_keith> mostly for me it has been engineering text books
[12:02] <vic_keith> their so expensive i worked it out that i was making $10/hour scanning them.
[12:02] *** netd has joined #active2
[12:02] <scott-van> heh... that's practically money in the bank ;-)
[12:02] <netd> hi. did you start already?
[12:03] * netd is sorry for being late
[12:03] <vic_keith> if you want you should check ot my ftp site... books music, etc.
[12:03] *** mike-away is now known as mike
[12:03] <mike> back now
[12:03] <vic_keith> hey hey
[12:04] <mike> hi all
[12:04] <netd> hoi :-)
[12:05] <mike> should we start?
[12:05] *** mike sets mode: +o vic_keith
[12:05] <GMD> hi... just curious about active 2.. :) don't have much tosay tho
[12:05] <vic_keith> lets begin
[12:05] <mike> hi; say anything you like, of course :)
[12:05] <mike> intros?
[12:05] <mike> mike warren from calgary, alberta
[12:06] <netd> ingo from bielefeld, germany
[12:06] <scott-van> scott nelson from vancouver
[12:06] <vic_keith> i'm keith, SWM, 6'4", like movies and sports.
[12:06] <scott-van> (hi ingo!)
[12:07] <scott-van> agenda item 1?
[12:07] <mike> okay, so any other agenda items?
[12:07] <mike> uhm, yea, okay, #1
[12:08] <mike> actually, should we get 2 and 3 done first? they seem quick...
[12:08] <mike> well, 2 anyway
[12:08] <vic_keith> added "user namespace" to the agenda.
[12:09] <netd> I'm in favor of that
[12:09] <mike> agenda 2: is ender still open for hacking active2? or is it now used for more important things?
[12:10] <scott-van> i think still open for hacking until we do the changeover... what's your feeling Keith?
[12:10] <vic_keith> the way it looks right now ender will be the new canada/west server and the rebuilt hiro will be the active2 server
[12:10] <vic_keith> but yeah its still open to hacking
[12:10] <vic_keith> until hiro is back
[12:11] <mike> okay, so it's not serving indy web content (yet)?
[12:11] <mike> when is changeover?
[12:11] <scott-van> we should have the changeover completed by monday and then hiro will be re-built as the active2-dev server next week
[12:11] <vic_keith> ender isn't live yet
[12:11] <vic_keith> the goal is to be there by friday
[12:11] <netd> would it be possible for me to still (or finally) use ender as a source for live, example data?
[12:11] <scott-van> netd: absolutely
[12:12] <vic_keith> so does that answer the question?
[12:12] <mike> vic_keith: yes. well i'll probably hold off hacking ender until after the changeover
[12:12] <mike> okay, so #2 closed
[12:12] <scott-van> will the twike stay on ender or will it be moved to hiro?
[12:13] <vic_keith> itll stay there
[12:13] <scott-van> kay
[12:13] <mike> so agenda 1: implementation framework
[12:13] <mike> i did some more hacking, and changed my previous example to be an example of my idea from last week
[12:13] <netd> mike: did you check out WebWare? (http://webware.sourceforge.net/)
[12:13] <mike> works okay;
[12:13] <mike> netd: no
[12:14] <netd> mike: I'm using it and am quite happy with it
[12:14] <netd> mike: it has a nice object-relational mapper (so, no stupid SQL code anymore)
[12:14] <mike> netd: okay, i'll check it out. what is it for, basically?
[12:15] <vic_keith> mike: do you have some code we could look at?
[12:15] <netd> mike: its basically a toolchest for web-developement
[12:15] <netd> mike: hooks into apache, has a constantly running server
[12:15] <mike> netd: just read the summary at their web site; sounds interesting, anyway
[12:15] <mike> netd: a single server, or one-per-httpd?
[12:15] <mike> vic_keith: yes, i do
[12:15] <netd> mike: a single server with threads
[12:16] <mike> http://meejah.homeip.net/~mike/ike-active2-hacking-may-8.tar.gz
[12:16] <netd> mike: what I like about it is that solves a couple of the annoyting things but
doesn't try to force you into some kind of framework. well, thats probably enough of that for now.
[12:16] <mike> netd: interesting. that would aid caching pre-parsed pythong objects (for example) quite usable...
[12:16] <mike> netd: how do you compare it to mod_python?
[12:17] <vic_keith> mike: thanks. i'll take a look
[12:17] <mike> vic_keith http://meejah.homeip.net/active2/article/foo/bar
[12:17] <mike> vic_keith is an example of it running
[12:17] <mike> i didn't think about the server-hook --> process protcol very much, as will probably be obvious...
[12:18] <netd> mike: I didn't use mod_python very much but I thought its kind of like mod_perl, right?
[12:18] <mike> netd don't know; never used mod_perl
[12:18] <mike> netd i think so, tho
[12:19] <netd> mike: well, its basically a persistent accelerator for scripts but the scripts still resemble normal CGI's, right?
[12:19] <mike> netd they can, but it also allows access to apache internals and other stuff like that
[12:19] <vic_keith> i think mod_perl and mod_python are the same in essence
[12:20] <vic_keith> scripting engines built into apache
[12:20] <mike> vic_keith yeah
[12:20] <vic_keith> mike: is your exasmple reading xml/newsml
[12:20] <mike> anyway, i was also interested if anyone else
[12:21] <mike> has done any hacking
[12:21] <mike> vic_keith no, not yet
[12:21] <mike> vic_keith it just puts the content into <pre> tags...
[12:21] <netd> mike: webware is on a higher level and has more application level toolkits built in
[12:21] <mike> i just wanted to see if my idea would work well
[12:21] <vic_keith> cool.
[12:22] <mike> would anyone object to a python-only first implementation?
[12:22] <scott-van> not me
[12:22] <vic_keith> nope
[12:22] <mike> my idea from last week was basically to facilitate non-python presentation-server
[12:22] <netd> is it decided already that we will re-implement everything from scratch?
[12:23] <mike> ...but we can always do that later
[12:23] <mike> netd: no, not necessarily
[12:23] <netd> mike: but thats what you're doing, currently, right?
[12:23] <mike> netd: for example, there are some
[12:23] <mike> newsml kits...
[12:23] <mike> netd: yes, more or less
[12:23] <netd> mike: do you have a pointer for the newsml kits?
[12:24] <vic_keith> what is the presentation-server written in?
[12:24] <mike> not off the top of my head; should be in logs+archives
[12:24] <mike> vic_keith the one i did is also in pythong
[12:24] <vic_keith> newsml.sourceforge.net
[12:24] <mike> vic_keith but it's just stdin/stdout, so...
[12:24] <vic_keith> ok
[12:24] <netd> keith: nothing there
[12:25] <mike> as far as real presentation servers, what ideas are there surrounding HTML-template type stuff?
[12:25] <vic_keith> http://newsml-toolkit.sourceforge.net
[12:25] <mike> obviously, this would be something whcih would be nice to use an existing package...
[12:25] <vic_keith> mike: yes
[12:27] <scott-van> netd: you had past objections to newsml... is this still the case?
[12:28] <netd> scott: well, not really
[12:28] <vic_keith> mike: what did you have in mind as far as templates go?
[12:28] <netd> scott: I just wonder wether Dublin Core would be enough, too
[12:29] <mike> no ideas, really
[12:29] <mike> vic_keith haven't really looked into it very much
[12:29] <mike> on my site, I use a custom python-hack to build all the pages
[12:29] <vic_keith> so is our python module taking in newsml and spitting out plain text?
[12:29] <mike> it works reasonably well; just enclose python in %...% in html-template...
[12:30] <netd> mike: did you check out www.cheetahtemplate.org?
[12:30] <mike> vic_keith it will give NewsML to the presentation-server subprocess (using my idea)
and get back HTML
[12:30] <mike> vic_keith (or actally whatever the user asks for)
[12:30] <mike> vic_keith i.e. the user might want application/x-pdf instead
[12:31] <mike> vic_keith in which case it would either be produced (if the presentation-server can
do it) or an error...(or some sane default)
[12:31] <scott-van> whoa... cheetahtemplate looks GOOD
[12:31] <vic_keith> yeah
[12:32] <mike> scott-van yeah, i just took a brief look. sounds good
[12:32] <mike> ...or at least interesteing
[12:32] <scott-van> heh... yah
[12:32] <vic_keith> mike: (sorry im not totally with you) would the pres-server subprocess be aware
of HTML templates?
[12:33] <mike> item for TODO: take a (serious) look at webware and/or cheetahtemplate
[12:33] <mike> vic_keith yes
[12:33] <mike> vic_keith the subprocess would be doing most of the real work
[12:33] <vic_keith> okay
[12:33] <mike> i'm still not convinced this is the best idea, though
[12:34] <scott-van> mike: what's the detail we're missing?
[12:34] <mike> scott-van: how do you mean?
[12:35] <vic_keith> would it make sense to have a different pres-server depending on the output
format (one for html, one for pdf)
[12:35] <mike> i meant i'm not sure that there aren't better ways (like maybe webware) to implement
[12:35] <mike> vic_keith yes, that's probably not a bad idea
[12:35] <mike> vic_keith the mod_python infrastructure could be told about which mime-types each
process knows about (for example)
[12:36] <mike> ...and potentially have a bunch of different processes it could call on...
[12:36] <scott-van> so for example a big video could be shoved through a bit-torrent process (or am i confused)?
[12:36] <vic_keith> i think that one existing system that could handle the templating/transformation is XSLT
[12:37] <mike> scott-van i was thinking that all the media-content would get served directly from
the content-servers (i.e. /content/filkey URLs)
[12:37] <netd> keith: transformation: yes, templating: no
[12:37] <mike> scott-van ...but there's no reason the content-server couldn't do that...
[12:38] <netd> keith: for the easy case (newsml->html), xslt is way too complicated. designer can't
stand it. for newsml->pdf, e.g., it could be cool
[12:38] <vic_keith> netd: okay... i agree its really complicated :)
[12:39] <vic_keith> i totally agree that the goal with templating is that it is easy to edit (resembles real HTML)
[12:39] <mike> vic_keith agreed
[12:41] <scott-van> so... where are we at?
[12:41] <vic_keith> so the html template has embedded calls to the pres-server, or uses an "object"
that contains the required content?
[12:42] <mike> let's not decide on templating yet
[12:42] <vic_keith> okay.
[12:42] <mike> vic_keith but yes, that's one good idea
[12:43] <mike> vic_keith ...more likely, the template has
[12:43] <mike> vic_keith sorry; wanted to abort that ;)
[12:43] <vic_keith> i like the idea of proceeding on a python (for now) - based pres-server
[12:44] <scott-van> gets my vote
[12:44] <mike> vic_keith me too; everyone here (i think) knows python and it's a great
proof-of-concept language (at least; it's good for applications too)
[12:44] * netd is totally in favor of that, too
[12:44] <mike> so, as far as agenda #1: more hacking needed :)
[12:44] <vic_keith> mike: i don't but i really want to learn it :)
[12:44] <mike> vic_keith sounds good :) it's a great language (IMO)
[12:45] <mike> so, another TODO item: explore concrete examples of how we might implement the server framework
[12:45] <mike> (that is, presentation-server framework)
[12:45] <netd> mike: how far along are you with the newsml stuff?
[12:45] <scott-van> mike: i don't (and I'm not really a programmer) but it has a great reputation
amongst those who count ;-)
[12:45] <netd> keith: newsml-toolkit seems java only...
[12:45] <mike> netd: haven't started; i was hacking out the general infrastructure
[12:45] <vic_keith> netd: yes - i just thought we could get some ideas from there.
[12:46] <mike> oh
[12:46] <mike> well, the presentation-server will need access to additional content when needed
[12:46] <mike> for example, it might need to get a comment to display, or grab some other article
for reference or whatever
[12:46] <mike> so any framework needs to let the presentation-server access the content-servers somehow
[12:47] <mike> that is, it can't just be given some NewsML and told to render HTML; it might need to
ask for more information while parsing the NewsML
[12:47] <vic_keith> mike: you mean access via http
[12:47] <mike> vic_keith yes, or via (for example) a Python call
[12:47] <mike> the latter might be a reason to stick with just python for now, as it might make
stuff like this easier to implement (at least for now)
[12:47] <netd> mike: well, xml becomes quite powerfull if you think of the *whole* site as one large xml document.
[12:48] <netd> mike: then everything is available :-) it requires an xml backend, such as
fourthoughts, though, and I don't know if we want that
[12:48] <mike> netd: that's true, but speed-wise, a presentation server might not need very much extra content
[12:48] <mike> netd and how should the presentatio-server framework decide which content to fetch...?
[12:49] <mike> netd oh, i see what you mean. XMLLink and such, potentially
[12:49] <mike> netd: IMO we should keep the content-servers as simple as humanly possible. a
couple weeks ago we decided they should just be Web servers (IIRC)
[12:49] <mike> obviously, they might have some modifications
[12:49] <netd> mike: I agree
[12:49] <mike> ...such as to talk to freenet
[12:50] <mike> but that's the idea; so talking to a content server for a particular extra bit of
info. will probably just be: http://utopia.indymedia.org/content/<filekey>
[12:50] <mike> so, can we conclude #1: python preferred; do more hacking this week to get some
examples of infrastructure in place
[12:51] <vic_keith> okay...
[12:51] <netd> mike: so, you are suggesting a model where the user accesses the presentation server
which then proceeds to retrieve the specified content from the content server?
[12:51] <mike> then, hopefully before next meeding
[12:51] <mike> netd: yes
[12:51] <netd> mike: keep me up-to-date on the newsml stuff, I'm very interested in that, too
[12:51] <mike> http://utopia.im.org/article/<filekey>
[12:51] <mike> for example, would engage the presentation-server to produce HTML for the NewsML
found in <fileky>
[12:52] <mike> whereas utopuia.im.org/content/<samefilekey>
[12:52] <mike> would return the raw NewsML
[12:52] <mike> netd: okay. i'll check out this webware stuff this week
[12:52] <vic_keith> me too.
[12:52] <mike> so, can we conclude #1?
[12:52] <vic_keith> ja
[12:52] <mike> attempt to produce examples for next week?
[12:53] <vic_keith> won't be here next week, but in 2 weeks for sure.
[12:53] * mike has to go soon, unfortunately...
[12:53] <scott-van> oh yeah... i won;t be here next week either.. two weeks for me
[12:53] <mike> so, what does agenda #3 mean?
[12:53] <mike> scott-van okay...
[12:53] <mike> is anyone else not available next week?
[12:54] <scott-van> i threw #3 in...
[12:54] <vic_keith> i think #3 is choosing a collaboration tool
[12:54] <netd> whats zeno?
[12:54] <mike> i don't have any problem with Wiki's. but yea, what's zeno
[12:54] <mike> how is it better than a wiki?
[12:55] <scott-van> http://zeno.berlios.de/
[12:55] <scott-van> a tool for assisting collaboration and decision-making
[12:56] <scott-van> maybe we can plan to talk about it at the next meeting??
[12:56] <mike> scott-van is it set up on ender? or anywhere?
[12:56] <mike> i'm just looking at the demo now, but hard to evaluate how it might work in practice...
[12:57] <vic_keith> mike: not yet
[12:57] <mike> looks somewhat interesting...
[12:57] <vic_keith> can always install it and try
[12:57] <scott-van> going to set it up on hiro i think
[12:57] <mike> i like the simplicity of Wikis...
[12:57] <vic_keith> it's somehow related to fraunhofer (the mp3 dudes)??
[12:57] <mike> scott-van okay, sounds good. how about post to list when it's up and we can all try
it out for a while and see how it compares to the wiki
[12:58] <scott-van> okay
[12:58] <netd> vic_keith: no, GMD -- the german national research institute -- recently got
"re-organized" to be in the same organization as fraunhofer. they are still way cooler than FhG.
[12:58] <vic_keith> so how bout we discuss user namespace on the list and then get to it next time?
[12:59] <scott-van> sounds good
[12:59] <vic_keith> we have a lot to go on for now i think... and its not really related.
[12:59] <mike> okay, so what's this user-namespace?
[12:59] <vic_keith> well, a global user-space, to go with our global file-space
[13:00] <vic_keith> for reputations, authentication, and whatnot
[13:00] <mike> oh, right
[13:00] <mike> can't we just use the global file-space?
[13:01] <mike> assign user-XML a file-key like anything else?
[13:01] <vic_keith> mike: possible
[13:01] <mike> then we just need a DTD for IMC-users
[13:01] <mike> ...this might not wor
[13:01] <mike> ...might not work very effectively (that is, quickly) for web-based logins to particular imcs, though
[13:02] <vic_keith> i was thinking something more "human", like "there is only one keith@victoria on
the indymedia network"
[13:02] <mike> it would work well with the idea of signed assertions to decide who wrote articles
(if they don't have PGP ro GPG keys)...
[13:02] <mike> vic_keith so there's a desire for meta-data about users, sort of?
[13:02] * netd has to leave, my room-mates are complaining about me blocking the phone
[13:02] <mike> netd: okay, thanks for the pointers
[13:03] <vic_keith> heh... netd email me if you need more db access.
[13:03] <mike> netd: see you next week
[13:03] <netd> mike: no problem :-)
[13:03] <scott-van> bye
[13:03] <netd> ok, see ya all!
[13:03] *** netd has quit IRC (Client Exit: Leaving)
[13:03] <vic_keith> bye
[13:03] <mike> vic_keith so what you're saying is there should be restrictions on user-names?
[13:03] <vic_keith> mike: yes
[13:03] <mike> vic_keith and how doe we enforce this in a distributed system?
[13:04] <vic_keith> mike: user logins when posting directly to imcs
[13:04] <vic_keith> all else is considered untrusted
[13:04] <vic_keith> unless another verification scheme can be found
[13:04] *** Anton has joined #active2
[13:04] <mike> vic_keith okay, so we're talking non-PKI users who have logged into an IMC?
[13:05] <Anton> hi all, sorry..
[13:05] <mike> hi, anton
[13:05] <vic_keith> yes, i think so.
[13:05] <mike> NP
[13:05] <vic_keith> hey anton
[13:05] <mike> okay, how about this: a user has XML data; it comes from a particular IMC; it is only
valid if it's signed by that IMC
[13:05] * Anton is "babysitting" his parents' cat..
[13:06] <vic_keith> many people have asked about having unique, trusted usernames.
[13:06] <mike> then, each IMC can gurantee uniqueness/work out log in stuff...
[13:06] <mike> vic_keith the username could be USER@IMC.indymedia.org, right?
[13:06] <mike> so it's only a valid username if IMC.indymedia.org has actually signed XML data for USER...
[13:07] <vic_keith> mike: yes... that's what i usually think would work
[13:07] <mike> okay, so the XML data has a filekey and an embedded IMC-signature
[13:08] <mike> and we can just distribute it as with any filekey'd data
[13:08] <vic_keith> yes that sounds good.
[13:08] <mike> vic_keith or, there could be a signed assertion from IMC saying USER is valid for IMC
[13:09] <vic_keith> hmm... i think this is something we should think about...
[13:09] <vic_keith> but i like the idea of rdf assertions
[13:09] <mike> i prefer the embedded signature idea; more compact
[13:09] <mike> ...plus the signature can verify the integrity of the user-data
[13:10] <mike> uhm. forgot to ask: was anyone logging this?
[13:10] <vic_keith> could this exist as a user-profile in XML
[13:10] <mike> vic_keith yes
[13:10] * mike has to got real-soon-now
[13:10] <vic_keith> that a user sends to an imc to prove identity or register
[13:10] <vic_keith> yes i am loggin
[13:10] <vic_keith> well discuss further on the list?
[13:11] <mike> vic_keith oh, i see what you mean
[13:11] <mike> vic_keith check out the AUTH project at freedevelopers.org
[13:11] <mike> sorry, dotgnu.org
[13:11] <mike> lots of interesting discussion in this realm
[13:11] <vic_keith> okay. id like to read more aobut this stuff.. complicated topic. important to
get it right.
[13:12] <mike> the IMC could give a user a XML-based certificate, though, yes, which they would then
present when submitting things
[13:12] <scott-van> kuro5hin has some good related articles on digital identity too
[13:12] <mike> I had some ideas on this...
[13:12] <mike> should be in the AUTH mailing list archives somewhere...
[13:12] <mike> scott-van true
[13:13] <vic_keith> i am really interested in user-profiles too. i want o hijack m$ .NET
[13:13] <vic_keith> but maybe that is for a different prohject? :)
[13:13] <mike> vic_keith ...then check out DotGNU.org, which aims to do exactly that
[13:13] <vic_keith> excellent
[13:14] <mike> (of course, DotGNU is interested in peole's freedoms, whereas .net isn't)
[13:14] <vic_keith> shall we call it a day? i'l post the logs.
[13:14] <mike> ;)
[13:14] <mike> vic_keith okay, sounds good.
[13:14] <mike> thatnks for the meeting
[13:14] <vic_keith> yeah, thanks again
[13:15] <vic_keith> later y'all
Session Close: Wed May 08 13:15:33 2002