User:Babbage Linden/Office Hours/2009 01 07
Jump to navigation
Jump to search
Transcript of Babbage Linden's office hours:
[8:10] | Babbage Linden: | hi everyone, happy new year!
|
[8:11] | WolfPup Lowenhar: | morning babbage
|
[8:11] | Babbage Linden: | sorry I'm a bit late
|
[8:11] | Nock Forager: | hello babbage. happy new year!
|
[8:11] | Fake Fitzgerald: | happy new year
|
[8:11] | Babbage Linden: | ok, i've pinged the scripters
|
[8:12] | Babbage Linden: | and the mono group
|
[8:12] | Babbage Linden: | so, how is everyone?
|
[8:12] | Babbage Linden: | any topics you'd like to talk about today?
|
[8:13] | Nock Forager: | Some changes on 1.25?
|
[8:13] | WolfPup Lowenhar: | i have an issue in jira about miss clicking
|
[8:13] | Ludo Merit: | afk brb
|
[8:13] | Nock Forager: | Prospero pinged about it today on preview grid group.
|
[8:14] | Babbage Linden: | prospero is the best person to talk to about 1.25
|
[8:14] | Babbage Linden: | i haven't been very involved in that release
|
[8:15] | Nock Forager: | He said some perfomance change on script may happen on 1.25. (So check it on preview before it came to the main.)
|
[8:15] | WarKirby Magojiro: | positive, or negative change.. ?
|
[8:15] | Nock Forager: | don't know.
|
[8:16] | Babbage Linden: | the only change i know of that would affect script performance is a slight change to the scheduler
|
[8:16] | Babbage Linden: | to fix a problem where some scripts could end up running for too long without yielding
|
[8:16] | Babbage Linden: | which caused time dilation problems
|
[8:16] | Babbage Linden: | so, it is worth checking your scripts
|
[8:16] | Babbage Linden: | but you shouldn't see any difference in performance
|
[8:17] | Nock Forager: | yup
|
[8:17] | Babbage Linden: | the only thing i've worked on in 1.25 is the script metric code
|
[8:18] | Babbage Linden: | which will tell us the script usage on a per parcel basis across the grid
|
[8:18] | WarKirby Magojiro: | will script metrics be available to everyone? only landowers ? only estate owners ?
|
[8:18] | Babbage Linden: | which we'll use for working out script limits
|
[8:18] | Babbage Linden: | intially it will be internal
|
[8:18] | Babbage Linden: | we're writing the data to logs and then parsing them to find out usage
|
[8:19] | Babbage Linden: | but we'll make the data available to everyone later
|
[8:19] | Babbage Linden: | so that everyone can compare their usage to proposed limits
|
[8:19] | Ludo Merit: | It would be very handy for estate owners. I want a low lag sim and don't quite know all the tricks to get it.
|
[8:19] | WarKirby Magojiro: | I mean, more specifically, will we be given acess to tools to measure the script time of our own scripts, as estate owners can now ?
|
[8:20] | Babbage Linden: | the metrics will initially be focused on memory usage
|
[8:20] | Babbage Linden: | as the big cause of lag is due to swapping
|
[8:20] | Babbage Linden: | due to memory exhaustion
|
[8:20] | Babbage Linden: | we're not measuring CPU usage of scripts atm
|
[8:21] | Babbage Linden: | the other script work we've been doing recently is on http-in
|
[8:22] | Babbage Linden: | which should be in 1.26
|
[8:22] | Babbage Linden: | we've been dicussing what to do about scripts which produce HTML
|
[8:23] | Babbage Linden: | which is potentially dangerous as a phishing opportunity
|
[8:23] | Babbage Linden: | as you can create HTML from a URL which includes lindenlab.com and so is more convincing than normal
|
[8:23] | Ludo Merit shudders.
| |
[8:23] | WarKirby Magojiro: | not really your fault if people abuse the tools they're given
|
[8:24] | jeaniesing Trilling: | is your problem if you have to prove, again and again, in court that it wasn't actually you....
|
[8:24] | Babbage Linden: | our initial approach was to set a text/plain mime type header
|
[8:24] | Babbage Linden: | but that wasn't respected by IE
|
[8:24] | Ludo Merit: | IE?
|
[8:24] | Babbage Linden: | which would render HTML content as HTML whether or not you specified text/plain as the content type
|
[8:24] | Babbage Linden: | internet explorer
|
[8:25] | Babbage Linden: | firefox and safari would render the HTML as text
|
[8:25] | Babbage Linden: | so you would see the markup
|
[8:25] | Babbage Linden: | but internet explorer would render it as HTML
|
[8:25] | Babbage Linden: | allowing you to create forms etc.
|
[8:25] | Ludo Merit: | Ah. I guess I am wise not to use IE
|
[8:25] | Babbage Linden: | and spoof linden web sites
|
[8:26] | Babbage Linden: | so, now we HTML escape data sent to IE
|
[8:26] | Babbage Linden: | making it appear the same as in firefox and safari
|
[8:26] | Babbage Linden: | i think this is the best approach
|
[8:26] | Babbage Linden: | it allows you to use web browsers to debug your web application
|
[8:27] | Babbage Linden: | but not to use HTTP-In to send fake linden web pages to browsers
|
[8:27] | Babbage Linden: | you can even cut and paste the HTML markup from the browser window to copy in to your application while debugging
|
[8:27] | Babbage Linden: | there were some calls to allow HTML generation from scripts
|
[8:28] | Babbage Linden: | but on balance we decided it was too dangerous
|
[8:28] | WarKirby Magojiro: | :/
|
[8:28] | Babbage Linden: | HTTP In will allow a performant and convinient replacement for XML-RPC and email
|
[8:28] | Babbage Linden: | and will allow round tripping of data using HTTML
|
[8:28] | Babbage Linden: | HTTP, sorry
|
[8:28] | Babbage Linden: | so, it's still very useful
|
[8:29] | Mephistopheles Thalheimer: | couldn't you just put /unauthorized/ in the URL?
|
[8:29] | Babbage Linden: | unfortunately not, because if you allow any HTML, an attacker can add javascript
|
[8:29] | Babbage Linden: | which can modify the rest of the page to hide warnings
|
[8:30] | Babbage Linden: | it's a tricky problem
|
[8:30] | Babbage Linden: | but i think we have a good solution now
|
[8:30] | Babbage Linden: | which makes HTTP-In useful but not dangerous
|
[8:31] | Casper Box: | add a header with a frame? ending with a noscript?
|
[8:31] | Babbage Linden: | that may be possible
|
[8:32] | Babbage Linden: | but we didn't want to allow HTML and then have to disable it
|
[8:32] | Casper Box: | i agree
|
[8:32] | Babbage Linden: | if there is lots of call for it, we can potentially look at it again
|
[8:32] | Babbage Linden: | at the moment we have a solution for our originally envisaged use case
|
[8:32] | Babbage Linden: | which is allowing 2 way communication with external web services using HTTP
|
[8:33] | Elbereth Witte: | will httprequest relults be capped less agressively in the rear future?
|
[8:33] | Elbereth Witte: | argh, spelling
|
[8:34] | Casper Box: | I find the cap now acceptable for most cases, 1 per sec is fairly resonable
|
[8:34] | Elbereth Witte: | I mean size, which is at around 2kbytes
|
[8:34] | Casper Box: | oh, that, yes
|
[8:34] | JS Uralia: | I thought it was 4kb before Mono
|
[8:34] | Elbereth Witte: | mono has raised teh bar to what we can process
|
[8:35] | JS Uralia: | yeah you should make that more
|
[8:35] | JS Uralia: | that should be a nice one-liner
|
[8:35] | Babbage Linden: | the 2KB was chosen as a reasonable amount of data for a 16KB script to process
|
[8:35] | Babbage Linden: | (I've spent lots of time watching scripts crash processing error reports from web services that were too big)
|
[8:36] | Babbage Linden: | so, yes it makes sense to raise the limit for mono
|
[8:36] | Babbage Linden: | although it will probably wait until we have script pools for memory
|
[8:36] | JS Uralia: | nobody will ever need peb pages longer than 640K
|
[8:36] | Babbage Linden: | once you have 500KB scripts, 2KB request limits will seem somewhat limiting
|
[8:37] | Elbereth Witte: | theres a threat of raising memory limits further?
|
[8:37] | Casper Box: | a HTTP_BODY_STARTAT wouldn't be bad, lets us grab it in chunks
|
[8:37] | Elbereth Witte: | agreed
|
[8:37] | Babbage Linden: | yes, chunked requests would be good
|
[8:37] | Babbage Linden: | (but time for scripting features is currently very limited due to many new strategies at the lab)
|
[8:38] | Babbage Linden: | i wouldn't expect any changes until after script memory limits
|
[8:38] | WarKirby Magojiro: | aww :(
|
[8:38] | JS Uralia: | Mono was script memory limits, to most scripters
|
[8:38] | WarKirby Magojiro: | so many features are still needed
|
[8:38] | Elbereth Witte: | any fuzzy guesstimate on that then?
|
[8:38] | WarKirby Magojiro still awaits the day we can change avatar clothing via script
| |
[8:39] | Aargle Zymurgy: | WarKirby, try using RLV. :-)
|
[8:39] | JS Uralia: | if you want it, put it in JIRA, then start a votes campaign, it's not hard to get random people to vote for jira issues, if you know them
|
[8:39] | Casper Box: | if you know them are they random?
|
[8:40] | WarKirby Magojiro: | I certainly do have a jira for it, and have for about a year now
|
[8:40] | Babbage Linden: | the top priorities from the last scripting survey were to have llGiveMoney and llGiveInventory with confirmation
|
[8:40] | JS Uralia: | then paste the url to like, Scripters Absurdly Large Group or something
|
[8:40] | Babbage Linden: | they are still probably top priorities after http-in and script limits
|
[8:40] | WarKirby Magojiro: | I remember that survey
|
[8:41] | Nock Forager: | confirmation, yes it's very needed function.
|
[8:41] | WarKirby Magojiro: | I don't recall any options for avatar related features like that
|
[8:41] | Babbage Linden: | but, we don't have any engineers working full time on scripting features at the moment
|
[8:41] | Neroelite Braun: | My vote is to have llDetectedGroup actually return a key :)
|
[8:41] | Elbereth Witte: | llGroupKey2Name() :-)
|
[8:41] | Babbage Linden: | and there are new engineering executives starting, so priorities and plans may change
|
[8:41] | Aargle Zymurgy: | where is that one Neroelite? I was just lamenting the lack of that yesterday
|
[8:42] | WarKirby Magojiro: | a year and a half old: http://jira.secondlife.com/browse/SVC-346
|
[8:42] | Neroelite Braun: | theres a whole jira thing on it with people arguing
|
[8:42] | Neroelite Braun: | yeah Iknow but still
|
[8:42] | JS Uralia wishes there were JIRAs for hiring more lindens https://wiki.secondlife.com/wiki/User_talk:Strife_Onizuka#Notecard_campaign_to_M_Linden
| |
[8:43] | Aargle Zymurgy: | people arguing in SL? wow.
|
[8:43] | Ludo Merit chuckles.
| |
[8:44] | Babbage Linden: | we've actually hired around 100 lindens in the last year, which is amazing growth
|
[8:44] | Aargle Zymurgy: | oh, that one deserves a real "wow"
|
[8:44] | WarKirby Magojiro: | what kind of lindens though ?
|
[8:44] | Babbage Linden: | but it takes a lot of interviewing and training time to bring that many people on board
|
[8:45] | JS Uralia: | 100 new hires, and 0 working on scripting features?
|
[8:45] | WarKirby Magojiro: | not 100 developers, I'd imagine..
|
[8:45] | Babbage Linden: | so a lot of us have been spending a lot of time on recruiting and on boarding
|
[8:45] | Babbage Linden: | the hires have been across the board
|
[8:45] | JS Uralia: | or by "0 full time" do you mean, how many full-time equivalent hours get on features?
|
[8:45] | Babbage Linden: | linden lab brighton has grown from just me to 14 people in the last 18 months
|
[8:46] | Babbage Linden: | which is huge growth and taken lots of time
|
[8:46] | Babbage Linden: | of those, 4 are developers and 4 web developers
|
[8:46] | WarKirby Magojiro: | nice :D
|
[8:46] | Aargle Zymurgy: | Brighton, England?
|
[8:46] | Babbage Linden: | i'm not sure what the demographics are across the company
|
[8:46] | Elbereth Witte: | seems like LL is gorging itsself on web developers
|
[8:46] | Babbage Linden: | yes, brighton, england
|
[8:47] | Babbage Linden: | lots of our internal services are now web services
|
[8:47] | Aargle Zymurgy: | hm... I wonder if Lamorna knows. you might get a visitor. :-)
|
[8:47] | Babbage Linden: | the project i'm working on at the moment is to move one of our C++ servers over to a web service
|
[8:47] | Babbage Linden: | so, it's not just the web site that requires web developers
|
[8:48] | JS Uralia: | all the more reason for larget httprequest buffers
|
[8:48] | Neroelite Braun: | hey babbage question for you
|
[8:48] | Neroelite Braun: | how do I get teh square rooot of a slice of pie
|
[8:48] | Neroelite Braun: | :)
|
[8:48] | Elbereth Witte: | an irrational question
|
[8:48] | Neroelite Braun: | being more indepth
|
[8:48] | Neroelite Braun: | its a slice of apple pie
|
[8:49] | Neroelite Braun: | I just thought I'd throw a joke in there.
|
[8:49] | Aargle Zymurgy: | ratio of the circumference to the diameter of an igloo is Eskimo Pi
|
[8:49] | Neroelite Braun: | HAHA
|
[8:49] | Neroelite Braun: | okay
|
[8:49] | WarKirby Magojiro: | it would be really nice to get a linden or two focusing on just something specific, like.... particles
|
[8:49] | Neroelite Braun: | I'm done
|
[8:49] | Neroelite Braun: | sorry
|
[8:49] | Aargle Zymurgy: | what more do you want in particles, WarKirby?
|
[8:49] | WarKirby Magojiro: | I found a few annoyances with particle flags yesterday. Apparently the linear flag is still in an alpha state, and undocumented
|
[8:50] | WarKirby Magojiro: | yet I find it utterly essential for something I'm working on just now, due to some odd behaviours it causes
|
[8:50] | Elbereth Witte: | its nearly all clientside I think
|
[8:50] | WarKirby Magojiro: | There's also particle glow, which would be nice to have. And perhaps some optisations for trails/beams. It's very hard to make a solid trail of particles that mvoe at any decent speed.
|
[8:51] | Babbage Linden: | ok, i need to head off in a couple of minutes
|
[8:51] | WarKirby Magojiro: | Also, every modern game in the world seems to have animated texture particles, which SL is lacking..
|
[8:51] | Babbage Linden: | any last thoughts or questions?
|
[8:51] | Aargle Zymurgy: | thanks for your time, Babbage. Sorry I missed the first half.
|
[8:51] | WarKirby Magojiro: | I think your right arm looks pretty nice babbage
|
[8:52] | Neroelite Braun: | I'm just happy knowwing linedns realy do exist. I totally just won a bet with my friends
|
[8:52] | Elbereth Witte: | llFloat2String() would be awesome, like, an accurate float casting
|
[8:52] | Aargle Zymurgy: | I just want to program in C#... nothing big.
|
[8:52] | Nock Forager: | oh upgraded?
|
[8:52] | Babbage Linden: | ah yes, i think there is a fix for float casting in Mono in 1.25
|
[8:52] | Babbage Linden: | and I want to program C# too Aargle
|
[8:52] | Nock Forager: | (right hand)
|
[8:52] | WarKirby Magojiro: | Oh, I wanted to ask about something
|
[8:52] | Babbage Linden: | I'm not going to stop lobbying for it ;-)
|
[8:52] | WarKirby Magojiro: | I'm seeing some odd issues with mono scripts
|
[8:53] | Aargle Zymurgy: | you've got a lot of backing, Babbage. :-)
|
[8:53] | WarKirby Magojiro: | where they occasionally crash out, with a pagefull of syntax in chat
|
[8:53] | WarKirby Magojiro: | stuff about reflections and parameters, I think..
|
[8:53] | Babbage Linden: | WarKirby, there is a bug for that in JIRA
|
[8:53] | Babbage Linden: | please add any useful information you can to that
|
[8:53] | WarKirby Magojiro will look and comment
| |
[8:53] | Babbage Linden: | we've fixed many causes of Mono stack traces
|
[8:53] | Babbage Linden: | but there may be some remaining
|
[8:53] | WarKirby Magojiro: | do you know which one it is ?
|
[8:54] | WarKirby Magojiro: | mono is a rather ubiquitous search term
|
[8:54] | Babbage Linden: | (the stack trace is a symptom of many problems, not just one bug)
|
[8:54] | WarKirby Magojiro: | doesn't narrow much down
|
[8:54] | Babbage Linden: | looking
|
[8:54] | Neroelite Braun: | babbage are you gunna beon on a regualr basis like this?
|
[8:54] | Neroelite Braun: | its nice being able to get questions answered
|
[8:54] | Babbage Linden: | yes, i have an office hour every wednesday at 8AM
|
[8:54] | jeaniesing Trilling: | Babbage, thank you for your time today... It is truly appreciated...
|
[8:55] | WarKirby Magojiro agrees
| |
[8:55] | Babbage Linden: | (i've just been off for a couple of weeks over the holidays)
|
[8:55] | Neroelite Braun: | thanks babbage. Much appreciated
|
[8:55] | WarKirby Magojiro: | An office hour about scripting is something that's been sorely needed
|
[8:55] | Neroelite Braun: | smart you having office hours when no one is really on :)
|
[8:55] | jeaniesing Trilling: | Would it be alright to forward that info to other scripting groups?
|
[8:55] | Aargle Zymurgy: | I'm always here. :-)
|
[8:55] | Babbage Linden: | it's at 4PM in Brighton ;-)
|
[8:55] | Neroelite Braun: | hahah oh
|
[8:55] | WolfPup Lowenhar: | babbage there is an issue with clicking in rc4
|
[8:55] | Mephistopheles Thalheimer: | only Australians know how to script anyway
|
[8:55] | Babbage Linden: | as late as I can make it and still have a hope of leaving at 5
|
[8:55] | Babbage Linden: | (which never happens)
|
[8:56] | Neroelite Braun: | make me a linden. I'll work the graveyeard shit
|
[8:56] | Neroelite Braun: | shift*
|
[8:56] | Babbage Linden: | the Mono stack trace bug is SVC-2908
|
[8:56] | Babbage Linden: | please add any new information there
|
[8:56] | WarKirby Magojiro: | Ah, I just found it, thank you
|
[8:56] | Babbage Linden: | http://jira.secondlife.com/browse/SVC-2908
|
[8:56] | Aargle Zymurgy: | nice typo, Nero. should have left it be. :-)
|
[8:56] | Babbage Linden: | ok, see you all next week
|
[8:56] | Neroelite Braun: | kk
|
[8:56] | Aargle Zymurgy: | bye babbage
|
[8:56] | Babbage Linden: | thanks for coming
|
[8:56] | Neroelite Braun: | bye
|
[8:56] | WarKirby Magojiro waves
| |
[8:57] | Neroelite Braun: | party in babbages office!
|
[8:57] | Nock Forager: | see you in next week thanks babbage. |