Difference between revisions of "User:Enus Linden/Office Hours/2008 July 25"

[9:29] Dale Innis: waves. :)
[9:29] Enus Linden: hi Dale
[9:29] Dale Innis: This beanbag keeps talking. :)
[9:29] Dale Innis: Hi!
[9:29] Tao Takashi: Hi
[9:29] Enus Linden: Hey Tao
[9:29] Enus Linden: howdy Sai
[9:29] Dale Innis: Hi various ppl!
[9:31] Enus Linden: I'll wait a couple of minutes to see if Infinity in particular rezzes in
[9:31] Bartholomew Kleiber: Hi all
[9:31] Dale Innis: Hello!
[9:31] Enus Linden: howdy!
[9:32] Locklainn Linden: hello hello
[9:32] Tomiko Magic: Sooorry to be on your head Batholomew >>.<<
[9:32] Tomiko Magic: I'll get off now...
[9:32] Dale Innis: haha
[9:32] Dale Innis: These things h appen
[9:32] Bartholomew Kleiber: lol
[9:32] Tomiko Magic: It's Enus' fault
[9:32] Bartholomew Kleiber: I thought we try a pyramid
[9:32] Tomiko Magic: lol
[9:33] Enus Linden: ok, so how about we begin...
[9:33] Enus Linden: Hi everyone, thanks for making it
[9:33] Dale Innis: seconded :)
[9:33] Tomiko Magic: :P
[9:34] Enus Linden: this will be an extension of the pyogp huddles that have been going on lately
[9:34] Enus Linden: and i'd like to reference in particular the hot topic of the day
[9:34] Enus Linden: https://lists.secondlife.com/pipermail/pyogp/2008-July/000106.html
[9:34] Enus Linden: So we've made architectural decisions, which are proving to be challenging in various ways
[9:35] Locklainn Linden: https://wiki.secondlife.com/wiki/Pyogp/Design_Decisions
[9:35] Enus Linden: buildout is so far not making life easier
[9:35] Locklainn Linden: I've done the last couple hours researching
[9:35] Enus Linden: yay Lock
[9:35] Locklainn Linden: and listed my findings on that page
[9:35] Tomiko Magic: wonders why the Wiki pages havn't been categorized Enus lol
[9:35] Locklainn Linden: they are
[9:35] Enus Linden: Mine are Tomiko, I'll add the tag to this one too
[9:35] Locklainn Linden: just not that one :)
[9:35] Tomiko Magic: : P
[9:36] Locklainn Linden: I made that 1 min ago haha
[9:36] Tomiko Magic: heh
[9:36] Tao Takashi: there should be some auto-tag module ;-)
[9:36] James Benedek: hola all
[9:36] Tomiko Magic: Poosie!
[9:36] Tomiko Magic: YAY : )
[9:36] Enus Linden: So, at a high level, I'm concerned that Tao is the single resourece familiar with much of the architecture we've put in place
[9:36] Enus Linden: Hi Mighty
[9:36] Enus Linden: we're in chat
[9:37] Dale Innis: No problem :)
[9:37] Enus Linden: no worries : )
[9:37] Ricken Flow: Hello everyone.
[9:37] Lillie Yifu: hihi everyone
[9:37] Dale Innis: Feel free to type along! :)
[9:37] Dale Innis: lols
[9:37] Tomiko Magic: lol
[9:37] Enus Linden: k, back to pyogp arch
[9:37] Enus Linden: pyogp needs to be accessible and maintainable down the road
[9:38] Enus Linden: and I'm not sure we're headed in that direction
[9:38] Locklainn Linden: as well as extensible and functional :)
[9:38] Enus Linden: Lock, you took the time to summarize on wiki, want to expound on the issues for me?
[9:38] Tomiko Magic: Big words FTW!
[9:38] Tomiko Magic: lol
[9:38] Locklainn Linden: sure
[9:38] Locklainn Linden: ok
[9:39] Locklainn Linden: so it seems there are 2 main problems
[9:39] Locklainn Linden: buildout
[9:39] Locklainn Linden: and zca
[9:39] Locklainn Linden: buildout in terms of getting the dev environment set up
[9:39] Locklainn Linden: and zca in terms of having "good" code
[9:39] Locklainn Linden: we have promises by Tao :) that buildout will function as we would like it to
[9:39] Locklainn Linden: so, as a quick overview,
[9:40] Locklainn Linden: we are using buildout to set up our development environment
[9:40] Locklainn Linden: so that we don't have to manually set up our paths, download our eggs and do our svn project checkouts
[9:40] Tess Linden: i spent several hours yesterday trying ot get buildout to work with Tao
[9:40] Tess Linden: buildout + bin/test
[9:40] Locklainn Linden: yea
[9:40] Locklainn Linden: it has problems now
[9:40] Locklainn Linden: but we have to work out those bugs
[9:41] Tao Takashi: all I can say is that these problems are sort of new to me as in the Zope/Plone world it usually works
[9:41] Locklainn Linden: note that it IS possible to dev without it
[9:41] Tao Takashi: but we can also develop without buildout
[9:41] Locklainn Linden: yea
[9:41] Locklainn Linden: it is then just the standard python way
[9:41] Tao Takashi: it's more or less a convenience thing (if it works)
[9:41] Locklainn Linden: setting up paths,
[9:41] Locklainn Linden: getting eggs
[9:41] Locklainn Linden: getting svn project dependencies
[9:41] Locklainn Linden: buildout does all that for us
[9:41] Tao Takashi: it's mostly easy_install zope.interface
[9:41] Tao Takashi: and so on
[9:41] Tao Takashi: I can write an alternative setup as well
[9:41] Tao Takashi: but I also added some note to the bug, Tess
[9:41] Tess Linden: please post the most simple and risk-free instructions on the wiki for the rest of us
[9:41] Locklainn Linden: may be a good idea
[9:42] Tao Takashi: because I would like to get rid of this bug if possible
[9:42] Tess Linden: thanks Tao, I'll take a look, but would prefer to get something that works first
[9:42] Tao Takashi: I understand this
[9:42] Locklainn Linden: is tess getting a different bug than me Tao?
[9:42] Locklainn Linden: the pyopenssl stuff?
[9:42] Tao Takashi: unfortunately I cannot test this on my machine as it works here
[9:42] Tao Takashi: Locklainn: Yes
[9:42] Enus Linden: buildout is proving itself cumbersome
[9:42] Enus Linden: we have diverse dev environments
[9:42] Tao Takashi: Locklainn: I am not sure if we can get rid of the pyopenssl dependancy except we don't install eventlet as an egg
[9:43] Enus Linden: and we tackle something were and find another thing there
[9:43] Locklainn Linden: right
[9:43] Tao Takashi: Enus: I know, the reason for inventing buildout though was to make it easier on those environments
[9:43] Enus Linden: i understand Tao
[9:43] Tao Takashi: seems that it does not live up to that promise yet
[9:43] Locklainn Linden: I think we just haven't learned how to configure it properly yet
[9:43] Locklainn Linden: it was built for this purpose,
[9:43] Tao Takashi: but the good side might be that I can fix these bugs now and report back to the zope community
[9:43] Locklainn Linden: I would say don't blame the code, blame the coder :)
[9:44] Enus Linden: but why spin wheels focusing on that when we can just go write lib code?
[9:44] Tao Takashi: well, it has some bugs obviously
[9:44] Locklainn Linden: unless its a code thing, but who knows
[9:44] Saijanai Kuhn: I've used it to install most things. I t seems to work fine. I think Lock's righ
[9:44] Tao Takashi: so, to sum it up: I will create some instructions
[9:44] Tao Takashi: as a fallback
[9:44] Locklainn Linden: alternative setup method?
[9:44] Tao Takashi: it will be some list of easy_install commands
[9:45] Tao Takashi: regarding windows though I am not the expert
[9:45] Tao Takashi: if some egg does not install maybe it would be good if somebody more familiar with windows can contact the mailing list for this package
[9:45] Tao Takashi: in case of eventlet maybe we can talk to Donovan and ask him to remove that dependancy for now
[9:46] Locklainn Linden: ...hmm
[9:46] Locklainn Linden: doesn't seem like the right fix
[9:46] Enus Linden: i just wonder if we can't eliminate buildout altogether
[9:46] Tess Linden: Lock: this is my jira: [1]
[9:46] Enus Linden: use svn externals
[9:46] Enus Linden: and a setup_path.py
[9:46] Locklainn Linden: thnx tess
[9:46] Enus Linden: we become less accessible externally
[9:46] Locklainn Linden: yea, that isa possibility enus
[9:47] Enus Linden: but can't we first get code written
[9:47] Enus Linden: and then explore buildout in a branch later
[9:47] Tess Linden: I would like to have a working set of instructions to run tests against the grid by July 30. Is that a reasonable goal?
[9:47] Enus Linden: when we can test it
[9:47] Enus Linden: and it's not blocking progress?
[9:47] Enus Linden: I think it is tess
[9:47] Tao Takashi: I will write the instructions after the meeting
[9:47] Locklainn Linden: awesome
[9:48] Enus Linden: tests against login and teleport yes tess?
[9:48] Tao Takashi: but it will pollute your python install then
[9:48] Tess Linden: and remember, theres several of us here... we can split up the tasks
[9:48] Locklainn Linden: Tao, doesn't that mean that everything will be installed to the system?
[9:48] Tao Takashi: usually I would prefer setting up a separate python installation
[9:48] Locklainn Linden: rather than to a dev environment?
[9:48] Tess Linden: we just need to coordinate and make sure we all agree on what we're doing
[9:48] Locklainn Linden: +1 Tess
[9:49] Tao Takashi: Locklainn: I will explain it with virtualenv then
[9:49] Tao Takashi: which sets up a complete separate sandbox with a copy of your python interpreter
[9:49] Locklainn Linden: well, that can be something we decide now
[9:50] Tess Linden: as a dev, I volunteer to test out whatever instructions you guys provide, as a sanity check
[9:50] Locklainn Linden: which way to do it
[9:50] Dale Innis: Or someone could just write a Perl script. :)
[9:50] Tao Takashi: we can of course also switch to Perl :)
[9:50] Tao Takashi: ok, so my proposal is to use virtualenv, then only this needs to be a global install
[9:50] Tess Linden: Tao: would virtual env mess up my python environment for other things on my mac though?
[9:51] Tao Takashi: Tess: no, the goal of it is not to do it
[9:51] Tao Takashi: it will copy your python interpreter and python libs to a separate directory
[9:51] Tao Takashi: with no connection to your original install
[9:51] Locklainn Linden: hmm
[9:51] Tao Takashi: in there you can then install packages locally
[9:51] Bartholomew Kleiber: like in plone, right?
[9:51] Locklainn Linden: can't we force it to use 2.5 then?
[9:51] Tess Linden: oh I get it now
[9:51] Tao Takashi: you can even try it out now:
[9:51] Tao Takashi: sudo easy_install virtualenv
[9:52] Tao Takashi: virtualenv somedirectory --no-site-packages
[9:52] Tao Takashi: and look inside that directory
[9:52] Tess Linden: I got an errorI got an error:
[9:52] Tess Linden: Error: There is a version conflict. We already have: zope.interface 3.3.0 but pyogp.lib.base 1.0dev-r885 requires 'zope.interface>=3.4'.
[9:52] Tess Linden: im afraid we're going down another rabbit hole
[9:52] Tao Takashi: well, Plone can be used with virtualenv but usually just buildout is used
[9:52] Tess Linden: (didnt mean to type that twice hehe)
[9:53] Tomiko Magic: lol
[9:53] Tao Takashi: well, let's check later on IRC I would say
[9:53] Tess Linden: We have very little time. Let's be as safe as possible
[9:53] Enus Linden: tess, what does as safe as possible look like to you?
[9:53] Tao Takashi: maybe we can make some skype call or so, might be easier
[9:53] Tess Linden: something that has worked before
[9:53] Tess Linden: its been a month since we started pyogp
[9:54] Enus Linden: standard python? build a package, use a setup file to add it to your path?
[9:54] Locklainn Linden: we've always used buildout
[9:54] Locklainn Linden: unless ur talking about enus' old code
[9:54] Tess Linden: just to be clear, i'm not suggesting that we scrap buildout
[9:54] Tao Takashi: ok, another option is to download all eggs needed and construct the path manually
[9:54] Tess Linden: I just want to begin with a simple thing, until buildout is ready
[9:54] Tao Takashi: Tess: can we do some voice call later? I think it would be faster
[9:55] Tess Linden: sure Tao
[9:55] Tao Takashi: ok, great. Then I might be able to find out what might be wrong
[9:55] Tao Takashi: even better would be ssh access but I guess that's not an option ;-)
[9:55] Enus Linden: :)
[9:55] Tess Linden: because I've tried the instructions on my mac *and* on Enus's station18 with the right python install
[9:56] Tao Takashi: ok, I would then write some instruction on how to set it up based on what I learn from Tess
[9:56] Enus Linden: ty Tao
[9:56] Tess Linden: thanks Tao
[9:56] Enus Linden: k, dev environment aside
[9:56] Tao Takashi: given I find a headset
[9:56] Enus Linden: (for now)
[9:56] Enus Linden: there is the issue of accessibillity, maintainability of the lib code itself
[9:57] Locklainn Linden: yep
[9:57] Locklainn Linden: and zca
[9:57] Enus Linden: given that we;ve chosen a framework that only one of us is able to work with comfortably at this point
[9:57] Enus Linden: yeah, zca
[9:57] Enus Linden: it has it's advantages for certain
[9:57] Enus Linden: and fulfills the stated requirement of being modular,
[9:58] Enus Linden: but for now, as we try to dev, all code goes through Tao
[9:58] Enus Linden: not a bad thing
[9:58] Enus Linden: but I know it's a time sink for all involved
[9:58] Tess Linden: the zca framework looks fine to me, but Ive had very little experience since I havent even gotten building+tests to work
[9:58] Enus Linden: and it would require much support as people start to adopt it down the road
[9:58] Enus Linden: imo
[9:58] Locklainn Linden: well,
[9:58] Locklainn Linden: zca is simpler than it is made out to be
[9:58] Locklainn Linden: all it is is a way to give Python interfaces
[9:58] Tess Linden: Lock: uve had a lot of experience with it, do you think it will be a lot to maintain?
[9:59] Locklainn Linden: like c++ and Java interfaces
[9:59] Locklainn Linden: it also gives us global registries
[9:59] Tao Takashi: well, I am trying to give as much support as I can for it
[9:59] Locklainn Linden: if you look at the wiki I wrote up,
[9:59] Enus Linden: and i thank you for it Tao :)
[9:59] Locklainn Linden: https://wiki.secondlife.com/wiki/Pyogp/Design_Decisions
[9:59] Tao Takashi: and I am also hoping to drag the grok author into this project
[9:59] Locklainn Linden: then there are the 3 main things I've found for the PRO ZCA arguments
[9:59] Locklainn Linden: so
[10:00] Locklainn Linden: everyone knows Python is dynamically typed
[10:00] Locklainn Linden: so you don't technically NEED interfaces
[10:00] Locklainn Linden: because anything can be an interface
[10:00] Tao Takashi: I wonder if most stuff in the library won't be very repetitive tasks anyway, like implementing another cap and implementing another udp packet
[10:00] Locklainn Linden: similarly, ZCA doesn't enforce this interfacing either
[10:00] Tao Takashi: and one can write recipe's for these tasks
[10:00] Locklainn Linden: it is a coding standard that developers agree to follow
[10:00] Saijanai Kuhn: sigh, that ultra lite intro to zca seems to have disappeared
[10:00] Locklainn Linden: it is the CONTRACT sort of
[10:01] Locklainn Linden: so
[10:01] Locklainn Linden: using the interfaces of ZCA allows us to use their global registry stuff
[10:01] Enus Linden: which sai?
[10:01] Tess Linden: should be in the wiki right?
[10:01] Tess Linden: anybody got a link?
[10:01] Locklainn Linden: so you can register an adapter, which takes an interface and adapts it to another interface (adding functionality)
[10:01] Bartholomew Kleiber: this one? [2]
[10:02] Tess Linden: thanks Bartholomew
[10:02] Locklainn Linden: and get that adapter anywhere in the code
[10:02] Locklainn Linden: this allows us to decouple our coding modules
[10:02] Locklainn Linden: so
[10:02] Locklainn Linden: in terms of maintaina
[10:02] Locklainn Linden: whoops
[10:02] Tomiko Magic: :P
[10:02] Locklainn Linden: so, it is a decision we have to make
[10:02] Locklainn Linden: if
[10:02] Locklainn Linden: we use ZCA
[10:03] Locklainn Linden: we get their global registry, which includes an event-subscriber system
[10:03] Locklainn Linden: interface adapters
[10:03] Locklainn Linden: and singleton utilties
[10:03] Tao Takashi: and Handlers
[10:03] Locklainn Linden: we enforce upon ourselves to code to interfaces (a great design principle)
[10:03] Tao Takashi: but that's similar
[10:03] Locklainn Linden: but if we don't use zca
[10:03] Locklainn Linden: we have to code all that ourselves
[10:04] Locklainn Linden: but we can still code to interfaces
[10:04] Locklainn Linden: as long as we explicitly state this somewhere in our coding guidelines
[10:04] Tao Takashi: well, prior to ZCA we used __implements__(ISomething) where ISomething was a normal class
[10:04] Locklainn Linden: aaaanad I'm spent :)
[10:04] Enus Linden: this sounds good to me, getting those coding guidelines written will be important
[10:05] Locklainn Linden: which sounds good
[10:05] Locklainn Linden: using zca, or not?
[10:05] Tao Takashi: thanks, Locklainn :)
[10:05] Enus Linden: using sounds OK imo
[10:05] Locklainn Linden: oh ok
[10:05] Tomiko Magic: Oki guys I gotta run nice meeting you all! Ciao!
[10:05] Enus Linden: i am/was just concerned abuot the burden we were putting on Tao
[10:05] Locklainn Linden: byte tomiko
[10:05] Enus Linden: bye tomiko
[10:05] James Benedek: bye
[10:05] Enus Linden: thanks for comin
[10:05] Locklainn Linden: there is a burden now
[10:05] Tao Takashi: well, now Locklainn also knows quite a bit about it :)
[10:06] Locklainn Linden: because the rest of us don't understand why its better
[10:06] Locklainn Linden: and how to use it
[10:06] Tomiko Magic: James get on Skype, I'm back later
[10:06] Locklainn Linden: once we do, it'll be much easier
[10:06] Tomiko Magic: Ciao!
[10:06] James Benedek: k
[10:06] Enus Linden: k
[10:06] Bartholomew Kleiber: bye tomiko
[10:06] Tao Takashi: well, I am happy to help out and I knew before that it means lots of explaining
[10:06] Locklainn Linden: I think the key is to know when to use what zca offers
[10:06] Enus Linden: so ZCA = good, and we can use it and maintain it, and tell others how to help with docs yes?
[10:06] Locklainn Linden: yes
[10:06] Enus Linden: yay!
[10:07] Tess Linden: I think if we assigned a task to each of us to write a test, we will all be hands on into the code, and provide much more support to Tao
[10:07] Enus Linden: So what are the actionable buildout items?
[10:07] Enus Linden: +1 Tess
[10:07] Locklainn Linden: what does that mean enus? :)
[10:07] Enus Linden: wishes Infinty were here to hand out test cases
[10:07] Tess Linden: Tao and I are going to spend 15 minutes or so on voice to see if we can work out our buildout issues
[10:07] Enus Linden: so buildout: it's annoying us
[10:07] Locklainn Linden: oh ok
[10:08] Saijanai Kuhn: isn't build out a deployment thing more than a svn thing?
[10:08] Tess Linden: I can make some suggestions on test cases
[10:08] Tao Takashi: yes, and I will write some basic instructions on how to setup a sandbox without it
[10:08] Enus Linden: Tao, you are going to take a look and provide updated docs? or did i misunderstand
[10:08] Enus Linden: ?
[10:08] Enus Linden: nice
[10:08] Tess Linden: here's the strawman link with the protocols:
[10:08] Tess Linden: https://wiki.secondlife.com/wiki/Second_Life_Login_API_Strawman
[10:08] Enus Linden: I'll help check those out..
[10:08] Tess Linden: oops sorry, thats the wrong link
[10:08] Enus Linden: Tess, we also have Infinity's list
[10:09] Locklainn Linden: where is that list enus?
[10:09] Tess Linden: https://wiki.secondlife.com/wiki/SLGOGP_Teleport_Strawman
[10:09] Tao Takashi: well, do we not working for LL don't have the problem that we cannot test rez_avatar etc. anyway?
[10:09] Enus Linden: i'm looking
[10:09] Tao Takashi: well, probably the opensim version
[10:09] Tess Linden: oh Tao, I have a solution to that. Theres an EC2 instance up of OpenSim
[10:09] Tess Linden: lemme get u the URL
[10:09] Tao Takashi: ok, I will install a version on my own here anyway
[10:10] Tao Takashi: (and write a buildout recipe which installs opensim ;-) )
[10:10] Tao Takashi: please post the URL somewhere :)
[10:11] Dale Innis: ooo, someone put an OpenSim up on EC2? I asked in Adam's weblog about that.
[10:11] Enus Linden: can't find infinity's list
[10:11] Tess Linden: [3] for now.. whump will post and send email to gridnauts
[10:11] Locklainn Linden: me either enus
[10:11] Dale Innis: kewl
[10:11] Tess Linden: zha hasnt released his patch since he's working on it daily
[10:12] Locklainn Linden: how do we sign up for the gridnauts list?
[10:12] Tao Takashi: I simply went to the listinfo page
[10:12] Tao Takashi: and signed up
[10:12] Tess Linden: mail.secondlife.com
[10:12] Tess Linden: didnt work..
[10:12] Lillie Yifu: https://lists.secondlife.com/cgi-bin/mailman/listinfo/gridnauts
[10:12] Locklainn Linden: nope
[10:12] Locklainn Linden: haha
[10:12] Tao Takashi: I hope he's releasing it soon :)
[10:12] Tess Linden: https://lists.secondlife.com/cgi-bin/mailman/listinfo
[10:12] Locklainn Linden: thanks Lillie
[10:12] Tess Linden: oh thanks Lillie
[10:13] Tao Takashi: or at least some in-development version
[10:13] Lillie Yifu: testing... lost in chat lag
[10:13] Tess Linden: sorry
[10:13] Enus Linden: so barring me finding infinity's wiki page, i'll find infinity and dole out test case writting via jira tasks
[10:14] Locklainn Linden: awesome
[10:14] Locklainn Linden: what about teleport
[10:14] Locklainn Linden: it seems to teleport there are some things we have to take care of first
[10:14] Enus Linden: like?
[10:14] Locklainn Linden: that is what i want to find out haha
[10:14] Locklainn Linden: I know teleport strawman lists some things
[10:14] Tao Takashi: [4]
[10:14] Locklainn Linden: but it seems there are otehr things, like udp messages
[10:14] Enus Linden: lemme sketch out a roadmap there
[10:14] Lillie Yifu: yes a patch version of zha's code is high on the list of things I would love to find in my Chirstmas stocking well before Christmas.
[10:15] Enus Linden: funny, searching OGP test case returned nothing useful :(
[10:15] Enus Linden: ty tao
[10:15] Tao Takashi: Lillie: Zha said that it might appear on forge today
[10:15] Tess Linden: ok, so Infinity already did extensive tests for authentication
[10:15] Enus Linden: [5]
[10:15] Lillie Yifu: kk
[10:16] Lillie Yifu: that would be glorious and wonderful and we would be very thankful for that
[10:16] Enus Linden: Infinity is working on that now tess
[10:16] Enus Linden: i'll ask how we can help her
[10:16] Tess Linden: ok we know what she's doing
[10:16] Tess Linden: we should split stuff up
[10:16] Tess Linden: do we have extensive tests for place_avatar or rez_avatar/place?
[10:16] Tao Takashi: what of course would be needed is error messages and formats when things fail
[10:16] Tao Takashi: like TOS etc.
[10:16] Tess Linden: Lock is working on message template stuff right
[10:17] Enus Linden: yes tess, that's locklainn's focus
[10:17] Locklainn Linden: tess, yess
[10:17] Locklainn Linden: yes*
[10:17] Locklainn Linden: builder is finished
[10:17] Locklainn Linden: going to start the reader today
[10:17] Tess Linden: ah, so Infinity's list is stuff we need to do but isnt done yet?
[10:17] Enus Linden: and place _avatar and rez_avatar/place are not tested yet
[10:17] Locklainn Linden: yea, she hasn't coded any tests yet, as far as I know
[10:17] Enus Linden: that's my understanding tess
[10:18] Enus Linden: let me take all these tasks and post them to jira and the roadmap
[10:18] Locklainn Linden: looking in the tests, there is only what was there the last time you and I looked tess
[10:18] Enus Linden: and we can sort them out in irc
[10:18] Tao Takashi: also good tests for the library itself
[10:18] Tess Linden: alright, theres 11 tests, how will we split them?
[10:18] Tao Takashi: I might add them to my mockup agent domain in there
[10:18] Tess Linden: +1 enus, pjira works for me
[10:19] Enus Linden: tess, point me to the test list and i'll feed pjira
[10:19] Locklainn Linden: [6]
[10:19] Enus Linden: ah those, kk
[10:20] Locklainn Linden: hmm
[10:20] Tess Linden: enus: I can add tests to pjira, just make a project for Authentication, Login, and Teleport (since these are the 3 pieces)
[10:20] Tess Linden: enus: add Inifinty's tests to Authentication
[10:20] Tess Linden: I will add more to Login and Teleport
[10:20] Enus Linden: make a project? or a jira meta issue?
[10:21] Tess Linden: however you see fit (not sure what it means in Jira)
[10:21] Enus Linden: :) will do
[10:21] Enus Linden: so per teleport
[10:21] Enus Linden: [7]
[10:21] Enus Linden: we can handle caps
[10:21] Enus Linden: we need to be able to handle AvatarMovementComplete
[10:21] Locklainn Linden: right
[10:22] Enus Linden: and TeleportFinished?
[10:22] Locklainn Linden: so, are we doing current SL teleport
[10:22] Locklainn Linden: ?
[10:22] Tao Takashi: and DinnerIsReady
[10:22] Enus Linden: which is the event queue yes tess?
[10:22] Tess Linden: We want to get rid of that path, but have not flushed it out completely
[10:22] Enus Linden: go eat Tao
[10:22] Enus Linden: :)
[10:22] Tess Linden: so no need to test after the seed cap step for now
[10:22] Tao Takashi: heh, just did :)
[10:23] Enus Linden: then we can accommodate tp in the libs with a little work and test up through the seedcap
[10:23] Enus Linden: i'll jira that as well
[10:23] Locklainn Linden: so, that brings me to another question. are we using pyogp just for OGP stuff, or for legacy as well?
[10:23] Tess Linden: Lock: OGP
[10:23] Tess Linden: py-ogp silly
[10:23] Enus Linden: tess, i'd hoped to support legacy if possible
[10:23] Locklainn Linden: hah well there was talk awhile ago about writing it for legacy too
[10:23] Enus Linden: with a focus on OGP
[10:24] Tess Linden: you mean, current SL login?
[10:24] Enus Linden: yeah
[10:24] Tess Linden: or specifically LegacyLogin?
[10:24] Saijanai Kuhn: don't see how we can test OGP without supporting legacy
[10:24] Locklainn Linden: well, everything really
[10:24] Enus Linden: UDP/caps/events are all the same
[10:24] Tao Takashi: we talked about having legacy in there right now as well and maybe moving parts to some legacy module as OGP matures
[10:24] Tess Linden: sure if you want, but I think we'd be too ambitious
[10:24] Saijanai Kuhn: well, you need UDP support to keep presence
[10:24] Tess Linden: Sai: we're using the event poll to keep presence
[10:25] Saijanai Kuhn: ah, did not know this was possible
[10:25] Enus Linden: OGP matures/we build libs to accommodate and test. background work can integrate legacy support
[10:25] Saijanai Kuhn: is this new with UDP?
[10:25] Tess Linden: if you stop polling the agent domain, it will disconnect you
[10:25] Enus Linden: it's not so much work imo
[10:25] Saijanai Kuhn: new with OGP?
[10:25] Tess Linden: enus: im fine with that if its not much work
[10:25] Tao Takashi: you mean reverse http or long poll?
[10:26] Tess Linden: Tao: those are implementation details of the same thing
[10:26] Tess Linden: Tao: so yes and yes :)
[10:26] Locklainn Linden: haha
[10:26] Tao Takashi: well, question would be what to implement ;-)
[10:26] Saijanai Kuhn: Tess, is this a recent development with OGP or has it worked that way with EQG always?
[10:26] Tess Linden: Tao: I think both work
[10:27] Enus Linden: whatevers is most time effective first Tao : )
[10:27] Tao Takashi: depends if you have something you can test against when implementing
[10:27] Tess Linden: Sai: we didnt have an agent domain before, so its sort of new?
[10:27] Saijanai Kuhn: ah, doh
[10:27] Tao Takashi: and where you also have access to the code and logfiles
[10:27] Tess Linden: Sai: dont remember if simulator disconnects you if you stop polling
[10:27] Locklainn Linden: I think sai means establishing presence, no?
[10:27] Saijanai Kuhn: right circuit code and all that
[10:28] Saijanai Kuhn: but yeah, was conflating sim presence and AD presence
[10:28] Enus Linden: k, times about up....
[10:29] Saijanai Kuhn: which leads to the queston: is udp handling required to keep sim presence?
[10:29] Enus Linden: i'll see y'all in irc in a bit
[10:29] Tao Takashi: k, Enus, thanks for hosting :)
[10:29] Enus Linden: thanks for the discussion y'all, and the patience :)
[10:29] Tess Linden: Sai: simulator doesnt disconnect you but stop talking to you via TCP/HTTP so you are essentially disconnected
[10:29] Tess Linden: UDP still works
[10:29] Locklainn Linden: thanks enus, and everyone else that came

Difference between revisions of "User:Enus Linden/Office Hours/2008 July 25"

Latest revision as of 10:36, 25 July 2008

Navigation menu

Search