User:Zero Linden/Office Hours/2008 Apr 22

[13:01] Rex Cronon: isn't there a more interesting topic? something related to vr? like for example sl?
[13:01] Arawn Spitteler: Chicory is a little poisonous
[13:01] Morgaine Dinova: Confectionary manufacturers should be made to pay a Fat Tax, to be taken out of profits alone, not passed to consumers.
[13:01] Kooky Jetaime: never happen
[13:02] Arawn Spitteler: Profit Taxes are passed to consumers, in the form of decreased investment. Everybody pays taxes.
[13:02] Kooky Jetaime: That will happen the day the oil companies actually have to pay proper taxes on their outragious prices
[13:02] Saijanai Kuhn: For those that missed the AW Groupies meeting this morning, we discussed the Agent Domain test last week and related stuff: https://wiki.secondlife.com/wiki/AWGroupies-2008-04-22
[13:02] Kooky Jetaime: checks gas prices.. oh, gas prices are up, looks like the oil companies profits will be even better this quarter
[13:03] Kooky Jetaime: wants to pull out his rifle and fix that issue but.. ugh.
[13:03] Arawn Spitteler: There's been talk of a Carbon Tax, to keep Petroleum competitive with Coal
[13:03] Morgaine Dinova: Cheers Sai. I started looking at that when you posted it earlier.
[13:03] Kooky Jetaime: Arawn - as long as we have an oil man in the chair, it'll never happen.
[13:03] Rex Cronon: there is ethanol available
[13:04] Arawn Spitteler: As long as we don't want to pay for our government, somebody else will.
[13:04] Rex Cronon: and hydrogen, although not a lot of cars that use it
[13:04] Saijanai Kuhn: Good Morning Teacher
[13:04] Saijanai Kuhn: needs to get a sound for that
[13:04] Harleen Gretzky: Hi Zero
[13:04] Zero Linden: hello all
[13:04] Morgaine Dinova: 'Morning Zero
[13:04] Aimee Trescothick: lol, did you bring an apple Sai?
[13:04] Saijanai Kuhn: chorus of little kids in first grade...
[13:05] Arawn Spitteler: Ethanol is a fraud. Coal could be processed into Methanol, but Ethanol starts with Agri-production, sending Food Prices Up, and winds up with more than its value in Processing
[13:05] Morgaine Dinova: Or 'afternoon even ;-)
[13:05] Zha Ewry: Hey Zero
[13:05] Rex Cronon: hi zero
[13:05] Goldie Katsu: Hey Zero
[13:05] Zero Linden: Uhm, okay - so this *ISN'T* the alternative fuels working group...
[13:05] Jamie Ondeko: lol
[13:05] Morgaine Dinova: That's not a bad topic .... virtual worlds and eating ;-)
[13:06] Arawn Spitteler: Well, we were worried about the energy probelm, but see you have your coffee, so we're fine
[13:06] Vent Sinatra: alternative fuel being coffe....?
[13:06] Rex Cronon: what. we were talking of new ways to powerup the sl servers. to make them run gree:)
[13:06] Goldie Katsu: looks at her banana
[13:06] Vent Sinatra: LOl Arawn!
[13:06] Mastorian Kingsford: ops thats why you didnt spawn chair was empty on my screen sorry haha
[13:06] Mastorian Kingsford: couch
[13:06] Zero Linden: ah yes, it's BYOC -- bring your own couch
[13:06] Vent Sinatra: wooops
[13:07] Vent Sinatra: that was my newbie imitation
[13:07] Arawn Spitteler: was actually thinking of ways to power his home SL Servers, using a mix of compressed air and surplus AC power
[13:07] Zero Linden: well, taking agenda topics....
[13:07] Mastorian Kingsford: i never leave home with out me trusty couch
[13:07] Rex Cronon: are they moded viewers that allow others to take over your computer?
[13:08] Vent Sinatra: lol mastorian
[13:08] Rex Cronon: there*
[13:08] Saijanai Kuhn: we talked about the Agent DOmain test this morning, Zero
[13:08] Aimee Trescothick: mixture of AC and compressed air sounds fun
[13:08] Saijanai Kuhn: https://wiki.secondlife.com/wiki/AWGroupies-2008-04-22
[13:08] Morgaine Dinova: 2-minute topic: is the nominal 2-year plan length of AWG work still thought to be 2 years? Are things on track?
[13:08] Zero Linden: Well - good agenda item 1) Milestones
[13:09] Zero Linden: agenda item 2) JSON
[13:09] Zero Linden: others?
[13:09] Morgaine Dinova: Oh cool, I like 2)
[13:09] Zero Linden: agenda item n) administrivia
[13:09] Teravus Ousley: likes JSON as long as we've got an object definition to serialize it too
[13:10] Zero Linden: I'm not hearing any more.....
[13:10] Zero Linden: SO
[13:10] Zero Linden: Welcome all to my office hours
[13:10] Morgaine Dinova: Tnx :-)
[13:10] Zero Linden: Speak freely and be welcome!
[13:11] Zero Linden: Now - item 1 - Milestones
[13:11] Tao Takashi: good evening
[13:11] Multi Gadget: v2.0.3b by Timeless Prototype, '/44 info'
[13:11] Morgaine Dinova: Hi Tao
[13:11] Teravus Ousley: Greetings Tao
[13:11] Rex Cronon: hi
[13:12] Zero Linden: Last Friday we had a bit of a milestone at LL
[13:12] Dahlia Trimble: new ceo?
[13:12] Zha Ewry: waves
[13:12] Zero Linden: This picture here
[13:12] Teravus Ousley: lol
[13:12] Morgaine Dinova: "Milestones" ... Is three an actual list of them, at least a vague mental one?
[13:12] Morgaine Dinova: there*
[13:12] Zero Linden: Which looks like a bunch of Ruths
[13:13] Zero Linden: Is a bunch of Ruths
[13:13] Zero Linden: But the key fact is that these Ruths are logged in to the grid named Aditi
[13:13] Saijanai Kuhn: so when you addign a web on a prim viewer so we don't have to wait for hte texture to rez?
[13:13] Dahlia Trimble: I can do that
[13:13] Zero Linden: but are standing in a region on the grid named Vaak
[13:14] Dahlia Trimble: oooo wormhole?
[13:14] Neas Bade: nice :)
[13:14] Zero Linden: In this case they are logged in via an Agent Domain run on the Aditi grid
[13:14] Saijanai Kuhn: more technically, run TO the Aditi grid?
[13:14] Zero Linden: but "RezAvatar"'d on a region in a different grid
[13:14] Rex Cronon: how did people knew that it was going to take place?
[13:15] Zero Linden: As I said, this was an internal LL milestone
[13:15] Morgaine Dinova: It's not a wormhole, it's just an exterior agent domain handling the av.
[13:15] Zero Linden: To do ti you needed a specail viewer that does the new login protocol and
[13:15] Mastorian Kingsford: my god man eat somthing your diet is really bad
[13:15] Zero Linden: know which hosts to point it all at
[13:16] Zero Linden: So, it doesn't look like much, but it demonstrated that the Agent Domain / Region Domain separation
[13:16] Morgaine Dinova: Very good
[13:16] Zero Linden: is going to be something we can get the current grid to get to
[13:16] Morgaine Dinova: What issues arose from the experiment?
[13:16] Zero Linden: And it shows you folks that there is real effort going on here at LL on this
[13:16] Zha Ewry: Severe ruthing
[13:17] Tao Takashi: rename it then to "A Ruth's Life" :)
[13:17] Zero Linden: Well, it showed what things the region you are in assumes it can get: your inventory, your shape, your clothes
[13:17] Saijanai Kuhn: thats just a matter of handing the client the right CAPs, isn't it?
[13:17] Morgaine Dinova: What aspect of the agent domain separation encourage ruthing?
[13:17] Zero Linden: Although, if you notice, Leyla Linden is wearing a jacket .... none of us could understand why!
[13:18] Zero Linden: Sai - well, for the client, yes
[13:18] Dahlia Trimble: did she change clothes after login?
[13:18] Tao Takashi: so that means that the agent domain was running on aditi?
[13:18] Zha Ewry: Well, we think we did figyre it out
[13:18] Zero Linden: so the clients were, for the most part, very very happy
[13:18] Zero Linden: the sim was a bit confused....
[13:18] Morgaine Dinova: LOL, fashion envy ;-)
[13:18] Neas Bade: I'd assume it was a cache issue, no?
[13:18] Zero Linden: Notice that not everyone's name is present
[13:18] Zero Linden: and we'd get IMs from ":"
[13:18] Zero Linden: You could build and attach - but not detach, only drop
[13:19] Tao Takashi: and it was handing the avatar over to a different region (on a different grid which probably is similar to different region domain)?
[13:19] Zero Linden: and not take object, but you could delete
[13:19] Zero Linden: Well, Tao
[13:19] Dahlia Trimble: inventory work?
[13:19] Saijanai Kuhn: gee sounds like my ypthing IM script on a good day
[13:19] Saijanai Kuhn: python IM script
[13:19] Zero Linden: this test didn't involve any region domain crossings -- only starting, and staying, in a region domain, that was wholly distinct from the agent domain
[13:19] Arawn Spitteler: You could attach, without Inventory?
[13:20] Tao Takashi: Zero: that's what I meant. agent domain on aditi, region domain is Vaak
[13:20] Zero Linden: Arawn - yes, it was very interesting, and several senior devs found it quite interesting to see what did and didn't work
[13:20] Zero Linden: Tao - yes
[13:20] Teravus Ousley: :D
[13:20] Zero Linden: and region crossing, within Vaak worked - another surprize
[13:20] Tao Takashi: so can we also at some point redo this experiment with some agent domain outside LL?
[13:20] Tao Takashi: e.g. one I write :)
[13:21] Mastorian Kingsford: was wondering what happened to db with client i seem to be 3500 items lighter and seems all the food is goon in my food folder
[13:21] Zero Linden: Tao - I think it will take the rest of the quarter to get code in shape for more rigorous trials (i.e.: ones involving outside servers)
[13:21] Mastorian Kingsford: ops gone
[13:21] Tao Takashi: because then I would feel motivated to work on one
[13:21] Zero Linden: but that is where we are heading
[13:21] Morgaine Dinova: Curious that it wasn't so much a test of expected s/w functionality, but more of a lab experiment :-)
[13:21] Tao Takashi: but very cool
[13:22] Saijanai Kuhn: seems to me test 1: agent domain => aditi; test 2: 2 agent domains => aditi ; test 3: some other grid
[13:22] Morgaine Dinova: Yep, definitely a milestone
[13:22] Tao Takashi: we should really define what "grid" means in AWG ;-)
[13:22] Zero Linden: well. to be honest, since it seems that OpenSim is closer to being a region domain
[13:22] Zha Ewry: Alkmost has one now
[13:22] Zha Ewry: The user code is nearly an agent domain
[13:22] Zero Linden: I think test two is our agent domain -> two region domains, one being external
[13:23] Neas Bade: nifty
[13:23] Tao Takashi: so if that works I should also be able to connect my AD to opensim
[13:23] Zero Linden: We have just gone through Q2 planning here at LL, but Tess (Studio Director of Studio Icehouse) was (and is) on vacation
[13:24] Zero Linden: I'd like to wait for her to get back, and let her tell you of her groups plans
[13:24] Zero Linden: Next week, I believe
[13:24] Zha Ewry: grins Good
[13:24] Morgaine Dinova: We could still talk about general project milestones goals though. Not dates, but expected points of functionality.
[13:24] Neas Bade: oh, that reminds me of a question, which maybe we can tack on the end
[13:25] Zero Linden: I just double checked, she'll be back next week
[13:25] Neas Bade: is there a documented set of services for -helperuri ?
[13:25] Zero Linden: Neas - I'll add that to the agenda...
[13:25] Neas Bade: great, thanks
[13:26] Dahlia Trimble: so some form of protocol compatability exists with opensim and LL's servers that would enable this (beyond compatability with the viewer)?
[13:26] Saijanai Kuhn: I've got a concern about AD and security that I'm confused about...
[13:26] Zero Linden: I think major milestones expected this year will include TP between region domains, inventory moved to agent domain, IM & Friends moved to agent domain
[13:26] Saijanai Kuhn: dalhlia its the new login protocol in e SLGOGP
[13:26] Saijanai Kuhn: and the rez avatar bit which may not be in there yet
[13:27] Morgaine Dinova: Quite a good list.
[13:27] Saijanai Kuhn: https://wiki.secondlife.com/wiki/User:Saijanai_Kuhn/Rez_Avatar_Capability
[13:27] Zero Linden: Morgaine - if we hit that list by Fall, I think we can all dance....
[13:27] Zha Ewry: checks her huddle
[13:27] Morgaine Dinova: Hehe
[13:27] Zha Ewry: Yep Dancing!
[13:27] Zero Linden: I'll even put dance balls her and we can hav ea party
[13:27] Electronics Aeon: Excuse me Zero, with the milestones expected, will a backup of these functions be server based?
[13:27] Tao Takashi: just invite us all to SF and we can dance there ;-)
[13:27] Xugu Madison: Zero, you make that list by fall, I'll certainly dance
[13:27] Vent Sinatra: lol tao
[13:28] Tao Takashi: btw, Call for Papers for EuroPython is open now. I hope Linden Lab submits something :)
[13:28] Zero Linden: E. Aeon - I'm not sure I understand - backup?
[13:28] Tao Takashi: (e.g. some AWG related topic is something I'd like)
[13:28] Tao Takashi: (or even some hackathon at the end of the conference might be nice)
[13:28] Morgaine Dinova: I guess that was fallback? Regression?
[13:28] Electronics Aeon: well say if my pc crashes and it's client side can i recover or backup from linden servers?
[13:28] Vent Sinatra: mmmm would love hackathon
[13:28] Pierre Wortsmann: hey!
[13:28] Zero Linden: E. Aeon - like the grid is now, all persistent state is stored in the servers
[13:29] Electronics Aeon: kk ty
[13:29] Tao Takashi: I just wonder how much of a hackathon could there be outside LL as they are probably working on closed source
[13:29] Zero Linden: Okay - agenda item 2: LLSD
[13:29] Zero Linden: and JSON
[13:29] Zero Linden: Now,
[13:30] Zero Linden: if you read the SLGOGP (yes yes yes... I hate the name too..... I'm hoping to get agreement to change it to just "Open Grid Protocol" )
[13:30] Zero Linden: if you read it carefully
[13:30] Tao Takashi: Open Grid Project would also be a good name btw :)
[13:31] Zero Linden: you'll see that I've tried hard to make LLSD be a *description* system for describing the structured data that are the reqest and response bodies
[13:31] Zero Linden: and left the serialization as a one of many
[13:31] Xugu Madison: i've kinda wondered about LLSD and XML; to me, it seems odd you're using XML to describe the data types, rather than what they mean...
[13:32] Zero Linden: Currently, the spec talks about (though does not define) an XML serialization
[13:32] Zero Linden: It is quite possible to have a JSON serialization
[13:32] Zero Linden: since the semantics of LLSD were carefully defined so that they meshed well with the kind of
[13:32] Morgaine Dinova: If it could be squeezed into JSON, that would be very efficient.
[13:32] Zero Linden: string-an'-numbers-an'-dicts-an'-arrays that many dynamic languages give you
[13:32] Zha Ewry: Squeeeze, isn't even the word
[13:33] Zha Ewry: fits well, I think is
[13:33] Teravus Ousley: heh, SL - AJAX client.
[13:33] Zha Ewry: Just need to define some things thst json doesn't
[13:33] Saijanai Kuhn: need a lot more proocols documented to port the client to AJAX
[13:33] Neas Bade: yen, one of the issues with that is that it maps very poorly to strong typed languages like C# and Java
[13:34] Allister Beerbaum: Zha: such as?
[13:34] Zha Ewry: nods
[13:34] Zha Ewry: "Formalists for URIs, and all"
[13:34] Zero Linden: Neas - well, we have a C++ implementation that supports all the features of dynamic languages in C++
[13:34] Morgaine Dinova: Hence the squeeze. :-) JSON is pretty basic, but that's part of what gives it great speed.
[13:34] Zero Linden: you can say things like
[13:34] Tao Takashi: my take on LLSD/JSON is usually that telling people "We use JSON" makes them feel more comfortable than learning a new name/concept
[13:34] Zero Linden: LLSD foo; foo["bob"
[13:34] Zero Linden: I'm trying to get those files released BSD
[13:35] Teravus Ousley: well, I've done some JSON work in .NET and it's simple and strongly typed if you know beforehand the object to serialize the data to
[13:35] Zero Linden: so that they can be freely used
[13:35] Saijanai Kuhn: object type = LLSD
[13:35] Tao Takashi: Teravus: and the protocol defines what is used for what
[13:35] Morgaine Dinova: Tao: that's true enough. JSON is so simple that it doesn't really have any dark corners, so yeah, it gives a comfort factor :-)
[13:35] Zero Linden: LLSD was designed so that if a host language didn't have support for a given data type, there were well defined round trip
[13:35] Zero Linden: conversions to and from strings
[13:35] Tao Takashi: my client hangs all30 secs for 1-5 seconds, I wonder if that's normal
[13:36] Zero Linden: so, for example, LLSD Dates, on a ssystem that doesn't have a date object
[13:36] Zero Linden: would represent these as ISO dates
[13:36] Tao Takashi: Morgaine: everything known to people already gives that :)
[13:36] Zero Linden: The C++ implementation automatically converts
[13:36] Rex Cronon: i was disconneted from sim, for some reason, tao
[13:36] Tao Takashi: they know how to use those libraries etc.
[13:36] Zero Linden: so
[13:36] Neas Bade: yeh, but it will end up being an issue. We'll need to get into the business of writing LLSD implementation for every language env that we want to use with it. I understand how well it maps to python, but we get JSON for free in almost any lang environment :)
[13:36] Allister Beerbaum: the nice thing about json is that it is very easy to implement
[13:37] Tao Takashi: it's well understood by most web devs
[13:37] Zero Linden: LLSD foo = "2008-03-01..."; LLDate d = foo;
[13:37] Pierre Wortsmann: thanks for the poopiebar moustache
[13:37] Zero Linden: does the right thing
[13:37] Morgaine Dinova: Tao: not really. You can work with C++ for a decade, and still find that comfort factor receding :-)
[13:37] Zero Linden: So,
[13:37] Tao Takashi: Morgaine: there might be a reason I never tried that ;-)
[13:37] Zero Linden: what is a problem is that JSON has no way to clearly encode binary
[13:38] Tao Takashi: so don't send binary ;-)
[13:38] Saijanai Kuhn: does xml?
[13:38] Zero Linden: well, for now, we have some binary
[13:38] Morgaine Dinova: base64
[13:38] Xugu Madison: XML uses base64, which isn't exactly ideal
[13:38] Zero Linden: The XML schema for LLSD does: <binary encoding="base64">akjdshflkajshdflkjashdflkahsdl</binary>
[13:38] Arawn Spitteler: Base 64 is just four char of Hex
[13:39] Zero Linden: Xugu - you'd prefer the far less supported Base96?
[13:39] Xugu Madison: My suggestion would be sending multipart messages, so binary is sent as an attachment rather than inline
[13:39] Tao Takashi: base42 ;-)
[13:39] Xugu Madison: Zero, I'd rather not have binary in my XML :)
[13:39] Morgaine Dinova: Heh, let's keep MIME out of this ;-)
[13:39] Zha Ewry: No On the wire MIMR
[13:39] Zha Ewry: MIME
[13:39] Zero Linden: Well, for better or worse, there are messages in the system today that have binary blobs
[13:40] Zha Ewry: No on the wire MIME!
[13:40] Tao Takashi: MIME otoh is not well understood by (web) devs ;-)
[13:40] Xugu Madison: I like MIME... and seriously, why not?
[13:40] Zero Linden: Well - let's be very very careful when we say "MIME" --
[13:40] Zero Linden: most of you, I bet, mean "MIME Types" --
[13:40] Zero Linden: Note carefully, for example, that HTTP does *NOT* transmit in MIME
[13:41] Zero Linden: the headers and body are expressly not MIME, though look awfully like it
[13:41] Xugu Madison: I mean something like multipart/form-data as a content encoding... although if MIME's too complicated (gyah header parsing), I'm sure we could think of a simpler multiple things in a stream format
[13:41] Tao Takashi: so what are those binary blobs for instance?
[13:42] Zero Linden: But, back to the issue - a JSON serialization of LLSD described requests and responses has only two issues
[13:42] Zero Linden: 1) Binary data.... which we could just say.... avoid, or not in JSON, or.....
[13:42] Zero Linden: 2) JSON requires all transmissions to be a map or an array
[13:42] Zero Linden: you can't use JSON to transmit a bare scalar value
[13:42] Saijanai Kuhn: well, you can't with LLSD either..
[13:43] Morgaine Dinova: Well so far there is just one missing requirement in JSON: binary types. I don't think we need to bring in all of MIME and especially the horrid multipart thing just to solve that, when something as simple as base64 would do the job.
[13:43] Zero Linden: You can: <llsd><string>Hello</string></llsd> is valid in the XML LLSD schema
[13:43] Zero Linden: and in the two other internally used serializations of LLSD
[13:43] Saijanai Kuhn: ah, OK. its still an llsd object though. But I see what you mean
[13:43] Zero Linden: Morgain- the problem with binary really is this:
[13:44] Zero Linden: well
[13:44] Zero Linden: I take it back
[13:44] Zero Linden: we'd just have to agree that binary data, converted to string in JSON, is base64'd
[13:44] Saijanai Kuhn: <llsd><<array>string>Hello</string></array></llsd> isnt' required
[13:44] Zero Linden: and that when you got a string from JSON, if you knew it was a binary blog, you'd un base64'd it
[13:44] Morgaine Dinova: Zero: the fact that you can't transmit a scalar and so need to encapsulate it as as 1-member structure isn't a problem though, because the overheads of the transmission vastly outweigh the cost of demarshalling a structure.
[13:44] Zero Linden: Sai - it isn't
[13:44] Tao Takashi: sounds good to me
[13:45] Xugu Madison: is happy with defining binary is base64, if it has to go inline
[13:45] Zero Linden: Morgain - I agree and I think we changed the spec at the last minute to rquire the top level of all requests and responses to be a map
[13:45] Zero Linden: if we dind't , we can
[13:45] Morgaine Dinova: Sounds fine to me.
[13:45] Teravus Ousley: I suppose we'd still have to consider endian issues?
[13:45] Zero Linden: Teravus - I don't think so - JSON is impervious to endianness
[13:45] Zero Linden: and binary blobs are always byte streams
[13:46] Saijanai Kuhn: Zero like EventQueueGet response?
[13:46] Morgaine Dinova: In any case, there is only One True Endianess.
[13:46] Morgaine Dinova: ducks
[13:46] Zero Linden: meaning - if there are multi-byte entities in there, the order is well defined with respect to the byte strem itself, not the endianess of the platform on either side
[13:47] Xugu Madison: Morgaine, ah, least significant bit in the middle to confuse people?
[13:47] Tao Takashi: so we all agree? :)
[13:47] Saijanai Kuhn: unlike SL packets...
[13:47] Zero Linden: Sai that is a legacy issue that should get fixed
[13:47] Zero Linden: it should be a map
[13:47] Zero Linden: Tao - sound like yes
[13:47] Saijanai Kuhn: yeah, I went through and corrected all fhte legacy login stuff to look like the real output. Not fun
[13:47] Morgaine Dinova: Xugu: hehe. Or the VAX ordering: 1 3 2 4
[13:47] Zero Linden: Now - we do have XML LLSD libraries for C++, PHP, Python and Perl
[13:48] Zero Linden: (though the later is probably old and out of date... but two weeks ago I worked redoing the PHP ones....)
[13:48] Xugu Madison: I can clean up and get permission to release our Java one if people want it, BTW
[13:48] Zero Linden: I will endevour to make those available
[13:48] Zero Linden: JSON is generally available
[13:48] Neas Bade: it would be nice to have a seperate project to at least link to implementations in many languages
[13:48] Zero Linden: and then all we need to do is adhere to content negotiation in HTTP and we're home free
[13:49] Morgaine Dinova: Literally many dozens of implementations. Most languages have several.
[13:49] Zero Linden: I agree - I'd also love them to have the same set of unit test data
[13:49] Zero Linden: I wrote the unit tests for C++ and PHP and they are rather extensive
[13:49] Zero Linden: (they test EVERY valid Unicode codepoint, for example)
[13:49] Tao Takashi: ok, so we just need to implement agent and region domains in all those languages then
[13:49] Tao Takashi: sounds easy
[13:50] Zero Linden: as my aunt in Brooklyn would say: "piece cake"
[13:50] Neas Bade: I bet you used klingon in those tests ;)
[13:50] Mastorian Kingsford: lol
[13:50] Saijanai Kuhn: not an official unicode member
[13:50] Zero Linden: Klingon is NOT officialy part of Unicode..... but those codepoints are valid!
[13:50] Neas Bade: testing high bit unicode is always fun
[13:50] Morgaine Dinova: Oh dear, I don't remember a Klingon port of JSON. I'll get to work on it straight away
[13:51] Zero Linden: JSON is Unicode - though I have to say, the rules for encoding detection are a bit arcane
[13:51] Teravus Ousley: [1] has many many of the implementations
[13:51] Zero Linden: I really wish they'd just have said "It's UTF-8. Done."
[13:51] Saijanai Kuhn: thinks Klingon is silly anyway. Should be a lot more complicated to write, like ARabic or Sanskrit
[13:51] Zero Linden: Okay - agenda 3: -helperuri
[13:51] Zero Linden: this one is easy - there are only two things that go thorugh that API:
[13:51] Tao Takashi: ok, I gotta go. cya later!
[13:51] Zero Linden: L$ purchase and Land purchase
[13:52] Zero Linden: later, Tao
[13:52] Rex Cronon: bye tao
[13:52] Tao Takashi: good meeting :)
[13:52] Xugu Madison: Seeya Tao!
[13:52] Morgaine Dinova: I've tested quite a range of the JSON implementations in serveral languages, and one of the interesting results was the huge range of performances --- easily 3 orders.
[13:52] Zero Linden: It is an ugly, pre capability API - and I think its days are numbered
[13:52] Zha Ewry: Good!
[13:52] Zero Linden: Morgaine - ditto for XML parsers!
[13:52] Neas Bade: Zero, ok, so how are those going to change over time?
[13:53] Morgaine Dinova: Zero: Hehe, not surprised. I've not tested those though :-)
[13:53] Zero Linden: I'm pretty sure that L$ will become an Agent domain message
[13:53] Saijanai Kuhn: that segueues (sp) into a security concern I have with the AD
[13:53] Zero Linden: Land purchase is unclear - is it really part of the fundimental feature set? IN this case it is the viewer pre-authorizing (and bying L$ if needed)
[13:54] Zero Linden: for account type and tier
[13:54] Zero Linden: So - perhaps it is also an Agent Domain message - but I wonder how we can standardize it....
[13:55] Morgaine Dinova: Land purchase isn't part of any fundamental feature set. PURCHASE is though, ie. exchange of anything for money, or maybe of anything for anythign else.
[13:55] Neas Bade: ok, I was hoping that somewhere in there was a way to redirect search in the client window to another server. I guess helperuri wasn't quite what I thought it was :)
[13:55] Saijanai Kuhn: wouldn't it be part of the intermediate "grid presence" part of rez avatar?
[13:55] Zero Linden: Right - so those will be in the Agent Domain protocols - as well as Region to Agent domain protocols for coordinating purchase involving vendors and/or objects
[13:56] Zero Linden: No, I'm pretty sure search doesn't go through the helperuri
[13:56] Arawn Spitteler: If party of hte first part is purchasing an axe or an acre from party of hte second part, then party of hte second part gives license to use axe or acre to party of hte first part, while the party of the first part extends license to draw funds, on the account of the party of hte first part, to the party of the second part...
[13:57] Morgaine Dinova: Hmmmm .... land isn't an object, but it probably should be!
[13:57] Saijanai Kuhn: which's office hours are on Thursday, Arawn
[13:57] Zero Linden: Plau =
[13:57] Zero Linden: Okay
[13:57] Neas Bade: zero, I didn't think so either. I think the search url is hard coded in the client xul files, but I was just hopeful :)
[13:57] Zero Linden: in my last remaining 3 min. (have a 2pm meeting)
[13:57] Arawn Spitteler: Anyone know what a Glow Rangle is?
[13:57] Zero Linden: So - no meeting this THursday - I'll be attending my kid's School play
[13:57] Lillie Yifu: Neils there is a debug setting that changes that.
[13:57] Morgaine Dinova: Sai had a security query re AD
[13:58] Saijanai Kuhn: Zero, a security issue maybe: passing off a avie from trusted agent domain to an untrusted is kool. But what about the otehr way?
[13:58] Aimee Trescothick: a lame attempt at griefing Arawn
[13:58] Zero Linden: And - still looking for people to take minutes - having a summary of the session is really nice
[13:58] Zero Linden: Sai - I don't think Avatars are ever passed from agent domain to agent domain
[13:58] Zero Linden: your avatar "lives" in the agent domain *you* trust
[13:59] Zha Ewry: frowns
[13:59] Saijanai Kuhn: hmmm... OK, was a possible scenario tess came up with concerning grid-specific rezzing issues.
[13:59] Zha Ewry: I don't think that works for some games, and restirctive envrionemnts
[13:59] Zero Linden: it is only presence on land that moves from region domain to region domain, trusted to untrusted and back again
[13:59] Zero Linden: Ah -
[13:59] Zero Linden: I do think that other agent domains can have "extended account info" or some such for your avatar
[13:59] Zero Linden: but I'd be pretty sure you need a "base" agent domain that you trust
[13:59] Zha Ewry: Ah
[14:00] Zero Linden: that hold you're inventory
[14:00] Zha Ewry: Probably yes
[14:00] Zero Linden: though not hte "games" inventory
[14:00] Zero Linden: exactly
[14:00] Zha Ewry: Well, one of them
[14:00] Saijanai Kuhn: see, if you DO allow that, your untrusted agent domian could do a man in the middle attack, posing as your client and keep tabs on what you're doing forever after
[14:00] Zha Ewry: I don't think you can expect all your inventory to be in one agent dfomain
[14:00] Zero Linden: though I wonder how much of that could be covered by just having per-agent-per-region-domain inventory
[14:00] Zero Linden: (when you are on this estate, you get these objects too....)
[14:00] Zero Linden: If so, that would simplify things greatly
[14:00] Saijanai Kuhn: goes back to needing a grid-specific presence before rez-avatar though
[14:01] Zero Linden: okay all
[14:01] Teravus Ousley: notes the time .. and has a nother meeting to attend as well
[14:01] Zero Linden: I've gotta run....
[14:01] Teravus Ousley: Take care :D
[14:01] Zero Linden: thank yo uall for coming
[14:01] Saijanai Kuhn: later zero. Thanks for doing this
[14:01] Bri Gufler: Thanks for having us Zero.
[14:01] Harleen Gretzky: bye Zero
[14:01] Morgaine Dinova: Zero: that brings up something that puzzled me from the start. Why doesn't the av live natively on your own machine? That's the only party you really trust. Why should the av "belong" to a World ISP" like LL? That's actually not very good. The "World ISP" should only be acting as non-authoritative big-iron for the av on the person's own equipment.
[14:01] Teravus Ousley: tc
[14:01] Rex Cronon: bye zero
[14:01] Dahlia Trimble: Bye Zero, and congrats on your test with the 2 grids
[14:01] Lillie Yifu: Really informative, thank you Zero.
[14:01] Tree Kyomoon: if someone could pass me the first 15 min or so I can post this
[14:01] Zero Linden: Morgaine- run an agent domain on your machine
[14:01] Saijanai Kuhn: Morgaine, goes back to ownership among other issues
[14:02] Xugu Madison: Seeya Zero!
[14:02] Morgaine Dinova: Zero: good answer :-)
[14:02] Zero Linden: but I, for one, log in from other machines
[14:02] Zha Ewry: Evening Zero..
[14:02] Zero Linden: think about e-mail and why one chooses IMAP over POP
[14:02] Saijanai Kuhn: Zero that's still an untrusted AD as far as SL is concerned
[14:02] Zero Linden: later

User:Zero Linden/Office Hours/2008 Apr 22

Navigation menu

Search