User:Which Linden/Office Hours/2009 Sep 17

• [11:03] Cel Edman: Saij what kind of blue bullet you are wearing?!
• [11:03] Saijanai Kuhn: and in fact...
• [11:03] Arthur Fermi: gave you Fermi Sandbox will close without help.
• [11:03] Cel Edman: .. talk about the bamboo
• [11:03] Latif Khalifa: hey Which
• [11:03] Which Linden: what up people!!!
• [11:03] Identity Euler:
• [11:03] Identity Euler: hi
• [11:03] Which Linden: nice unicode
• [11:04] Saijanai Kuhn: its a nboleman's hat from the Heian era (check out t-shirt)
• [11:04] Saijanai Kuhn: Dahlia made it for me
• [11:04] Cel Edman: kk
• [11:05] Which Linden: so hey, quick poll for y'all
• [11:05] Which Linden: does anyone here care about the "Give Inventory" entries in the transaction log?
• [11:05] Latif Khalifa: not really
• [11:06] Chilli Cao: yeah like twice a year.. when im trying to find something
• [11:06] Chilli Cao: say that i know someone gave me something that i dont know the name
• [11:06] Which Linden: they only seem to apply to inventory transferred directly
• [11:06] Saijanai Kuhn: [1]
• [11:07] Saijanai Kuhn: is there a different packet for that then for imed inventory?
• [11:07] Cel Edman: 20:00 pm here ^^
• [11:07] Latif Khalifa: Cel, same here
• [11:07] Which Linden: I mean it applies to inventory dragged from one inventory to another av, but not to inventory transferred from objects
• [11:07] Saijanai Kuhn: ah OK
• [11:08] Which Linden: so if you give inventory to someone who's offline it writes the transaction log first and then stores the IM
• [11:08] Which Linden: so even if they decline the inventory or don't get it, the log remains
• [11:08] Latif Khalifa: Which, I personally never had use of it, i guess its not annoying
• [11:08] Chilli Cao: its proof that someone sent something
• [11:08] Saijanai Kuhn: there's no confirmation log?
• [11:08] Latif Khalifa: and "decline" really is only client side
• [11:08] Latif Khalifa: for agent -> agent transfers
• [11:09] Latif Khalifa: in reallity in means "tell the guy i declined, and move it to my trash"
• [11:09] Cel Edman: ((wondering when there is a message cap. like recently I joined an art group that sends out a lot of messages about 5+ each day, and now i`m thinking to leave this group because it could cap out other messages.
• [11:10] Chilli Cao: notices?
• [11:10] Identity Euler: you can turn off messages
• [11:10] Identity Euler: notices*
• [11:10] Latif Khalifa: and message cap is really evil since you lose all inventory offers from objects once you hit the msg cap
• [11:10] Which Linden: Yeah I guess when the inventory is offered initially it actually goes into your inventory somewhere, and the "decline" just deletes it
• [11:11] Which Linden: I think anyhow
• [11:11] Cel Edman: most of that goes to the dust bin
• [11:11] Latif Khalifa: default is accept from agents, so if you don't accept and say logout it will still be in your inv
• [11:11] Which Linden: The accept button is mostly a no-op :-)
• [11:11] Latif Khalifa: for agent - agent
• [11:12] Latif Khalifa: object - agent is a different story entirely
• [11:12] Chilli Cao: the "give inventory" in transaction log you asked.. you meant only the time where object llGivesInventory ? or any
• [11:13] Cel Edman: Well to be honest, if people want to get back to you and it wont work on the first try, try again.
• [11:13] Which Linden: Latif: right, good distinction
• [11:13] Which Linden: Chilili Cao: if you log into secondlife.com and go to "Transaction History" there are a bunch of logs there....some of them say "Give Inventory"
• [11:14] Which Linden: llGiveInventory doesn't generate these logs
• [11:14] Which Linden: only someone dragging stuff from their inventory onto you or your profile
• [11:14] Which Linden: lemme show you
• [11:15] Chilli Cao: i see.. personaly i would prefer them there.. just for sake of completeness.. if all inventory input should be logged
• [11:15] Chilli Cao: accepted your inventory offer.
• [11:15] Latif Khalifa: why do you ask? want to kill those from the transaction log?
• [11:15] Which Linden: Yeah I agree that ideally they'd remain
• [11:15] Which Linden: But it's work to maintain features, and it might save us some work to kill this one
• [11:15] Which Linden: Yes, Latif, I'm debating killing them :-)
• [11:16] Which Linden: so ok! that was useful to know
• [11:16] Cel Edman: Which Linden> What is your current project you're working on to improve/make things easier now, If I would ask?
• [11:16] Latif Khalifa: i wouldn't lose sleep if they weren't there ;)
• [11:16] Which Linden: I'm obviously not going to make my decision right away but that was some valuable feedback, really appreciate it
• [11:17] Chilli Cao: i nowdays that keep chat.txt go there and search "which linden gave you"
• [11:17] Chilli Cao: to find an item that i know who gave it to me but not its name
• [11:17] Identity Euler: yes there is no name of item right?
• [11:18] Chilli Cao: but thats 120mb textfile already
• [11:18] Chilli Cao: there is a name, i just dont remember it
• [11:18] Chilli Cao: maybe its japanese name
• [11:18] Identity Euler: ok
• [11:18] Chilli Cao: i was looking for a watermelon piece the other day
• [11:18] Chilli Cao: SUIKA in japanese
• [11:18] Which Linden: Cel: my current project is to move L$ onto its own database, separate from the main database
• [11:18] Which Linden: that will hopefully make both databases happier
• [11:19] Chilli Cao: but since i remembered who gave it to me.. i could go and search for that name in my transaction log
• [11:20] Chilli Cao: anyway its a rare case isnt it.. for the minority that are like me.. chat.txt saves the day
• [11:20] Cel Edman: Well for me micro payments are the future for me, atm only paypall seems king in general, but they ask like 35 eurocents for every transaction. I guess maybe google goes next in combination with addwords. but SL micropayments is the best thing of sl.
• [11:21] Which Linden: Totally. Which is why they should be fast and reliable. :-)
• [11:21] Chilli Cao: but when you right click pay for an item.. then vendor gives you inventory .. that should remain logged no?
• [11:22] Which Linden: That is not currently logged
• [11:22] Chilli Cao: oh
• [11:22] Which Linden: The payment part of that is currently logged, but the inventory part is not
• [11:22] Which Linden: Which is a bummer
• [11:22] Chilli Cao: how old of a transaction log can i get from my dashboard?
• [11:22] Which Linden: 30 days
• [11:22] Which Linden: maybe 31
• [11:22] Chilli Cao: i see
• [11:24] Cel Edman: I try to make backups every month.
• [11:24] Latif Khalifa: yes, that one would be useful, a transaction log (separate from $L) that logs your object inventory offers and if they were accepted or not
• [11:24] Which Linden: Does anyone want to resume looking at the source code to try and make progress on SVC-4182?
• [11:25] Latif Khalifa: in 30min? i don't think it can be done in that time
• [11:25] Which Linden: darling brody's latest comment that this is a years-old bug is....I guess predictable
• [11:25] Which Linden: No I don't think we'll find the bug in 30 min
• [11:25] Which Linden:  :-)
• [11:27] Cel Edman: Which make a working link for SVC-4182.
• [11:28] Which Linden: [2]
• [11:28] Which Linden: We need a bot that generates links
• [11:28] Latif Khalifa: there is a jira helper object
• [11:28] Which Linden: oooooo
• [11:28] Latif Khalifa: i'll ask dahlia, or moon for it
• [11:30] Latif Khalifa: SVC-4182
• [11:30] JIRA-helper: http//jira.secondlife.com/browse/SVC-4182:
• [#SVC-4182] being deducted: twice when paying in-world objects
• [11:30] Latif Khalifa: :D
• [11:30] Latif Khalifa: gave you JIRA-helper.
• [11:31] Latif Khalifa: now it will link to jiras automatically hehe
• [11:31] Which Linden: Thanks!
• [11:31] Latif Khalifa: SVC-4444
• [11:31] JIRA-helper: http//jira.secondlife.com/browse/SVC-4444:
• [#SVC-4444] become fully: permissive when rezzed under certain conditions. - with Repro
• [11:31] Latif Khalifa: hehe
• [11:32] Which Linden: There we go, behind the couch
• [11:32] Morgaine Dinova: Hi all :-)
• [11:32] Latif Khalifa: hi Morg
• [11:33] Latif Khalifa: which, so $L transacion cause significant db load?
• [11:33] Which Linden: MISC-2540
• [11:33] JIRA-helper: http//jira.secondlife.com/browse/MISC-2540:
• [#MISC-2540] Life would: rock harder with the mano cornuta hand gesture
• [11:33] Morgaine Dinova: Busy with relatives. :-) What's today's topic?
• [11:33] Latif Khalifa: hehe
• [11:33] Which Linden: Latif: yes, significant write load
• [11:34] Which Linden: Morgaine: well, was hoping to continue looking at SVC-4182, but not wedded to that, since we're not likely to make a ton of progress
• [11:34] JIRA-helper: http//jira.secondlife.com/browse/SVC-4182:
• [#SVC-4182] being deducted: twice when paying in-world objects
• [11:34] Which Linden: Also this JIRA helper is the mutt's.
• [11:34] Latif Khalifa: should be easy enough to remove from the main db, cannot imagine that its too interconnected
• [11:34] Latif Khalifa: it only needs user db right
• [11:35] Cel Edman: I guess, I better make the tools next. before the market is ready for it.
• [11:35] Latif Khalifa: or do you write a copy in each log entry?
• [11:36] Latif Khalifa: has customer on IM complaining about double payment right now
• [11:36] Latif Khalifa: which if you want to lookup the log
• [11:37] Which Linden: Latif: yeah can I get a copy of the viewer log?
• [11:37] Latif Khalifa: erm, i don't think i can convice her to do that
• [11:38] Morgaine Dinova: Not really my area. Making things happen only once isn't exactly pushing the limits of technology :P
• [11:38] Which Linden: Um....so should I talk to her with my Lindeny powes?
• [11:38] Which Linden: Morgaine: it sounds like you've not heard of the two-generals problem then
• [11:39] Morgaine Dinova: Generals are bad, in general ;-)
• [11:39] Which Linden: Generally I'd agree with that :-)
• [11:39] Morgaine Dinova: Hehe
• [11:39] Latif Khalifa: 1925808336 and 1925808337 are transaction numbers
• [11:40] Chilli Cao: are they on wifi
• [11:41] Which Linden: Latif: ok checking
• [11:41] Latif Khalifa: sent the customer name via IM which
• [11:42] Which Linden: Thank you, will follow up
• [11:42] Cel Edman: Anyway, a global micro-payment system around the world. could change the world I think and feel.
• [11:43] Latif Khalifa: google is all over it
• [11:43] Aimee Trescothick: VWR-15549 is the one we found last week btw
• [11:43] JIRA-helper: http//jira.secondlife.com/browse/VWR-15549:
• [#VWR-15549] fast double-click: on an object with the "Pay Object" click action set, can result in multiple super-imposed payment floaters
• [11:43] Latif Khalifa: but it would be funky if Linden Lab would allow web sites to charge in Lindens
• [11:44] Saijanai Kuhn: according to the guy who advises Congress on virtual worlds, if L$ starts getting used to pay for real world services, it becomes taxable imediately instead of during cash out
• [11:44] Which Linden: Aimee: awesome, thank you
• [11:45] Which Linden: Sai: who's that?
• [11:45] Aimee Trescothick:  :)
• [11:45] Saijanai Kuhn: can't remmeber his name. He spoke atMetanomics last year I think
• [11:45] Saijanai Kuhn: big fan of SL and WoW
• [11:47] Which Linden: Oh ok. I wonder if that's actually the case
• [11:48] Morgaine Dinova: Wasn't this issue essentially solved by the work on CHTTP? Charging for a given sale should be an idempotent operation.
• [11:48] Which Linden: Morgaine: yes, apparently the viewer is sending two unique requests
• [11:49] Which Linden: It's probably a UI problem
• [11:49] Saijanai Kuhn: [3]
• [11:49] Which Linden: Entropia became a bank
• [11:51] Morgaine Dinova: But if the viewer is sending two unique requests to buy the same item, the second one should fail.
• [11:51] Latif Khalifa: hey!
• [11:51] Latif Khalifa: no lol, people buy more that one item all the time
• [11:51] Which Linden: Morgaine: that would be the case if the semantics of the transaction were "buy this object" but in this case it's "pay this vendor and see what happens"
• [11:51] Morgaine Dinova: No no Latif, the SAME item, not multiple items of the same kind
• [11:51] Latif Khalifa: besides, that would be hack to mask a real problem
• [11:52] Which Linden: Agreed that if we changed the semantics so that the viewer was requesting to buy a *specific* object the problem would be solved
• [11:52] Which Linden: but since the semantics are of buying a *copy* of an object it's not so easy
• [11:52] Latif Khalifa: Morgaine, paying strippers small amounts with great frequency is a common case :P
• [11:52] Latif Khalifa: don't break those tipjars :P
• [11:53] Which Linden: Yes, good example Latif
• [11:53] Morgaine Dinova: That's interesting, a semantic difference between buying an item and paying for a sevice.
• [11:53] Which Linden: At some point it gets irreducible: the problem becomes "pay as many times as the user wanted to pay, regardless of what their fingers did"
• [11:54] Latif Khalifa: there are some dragons in that floater, we have seen it the last time
• [11:54] Morgaine Dinova: I think you have to narrow it down to payment for a unique object, otherwise it's a recipe for accounts being drained when things go wrong.
• [11:55] Which Linden: Morgaine: I think you're right in a sense, but it's annoying to implement
• [11:55] Morgaine Dinova: Eg. a "tip" could be payment for a reciept for services covering a certain span of time.
• [11:55] Morgaine Dinova: And the unique items is the receipt
• [11:55] Which Linden: Because then the process of "tipping" someone becomes a process of "ask recipient to generate unique 'object', buy it, repeat"
• [11:56] Which Linden: Yeah we're in agreement about how it would work
• [11:56] Which Linden: What if you accidentally generate two receipts though????
• [11:56] Which Linden: O.O
• [11:57] Which Linden: It's turtles all the way down
• [11:57] Morgaine Dinova: There's no mouse or keybounce server-side, luckily :-)
• [11:58] Which Linden: This is true. But it's not clear this problem is caused by keybounce
• [11:58] Freemason Magic: i have to go thanks for the information
• [11:59] Which Linden: I mean, technically the UI elements should disable themselves upon receiving the click event, so UI elements should not ever suffer from multi-clocks
• [11:59] Which Linden: *clicks
• [11:59] Which Linden: Perhaps that guarantee is being violated here, which would be a problem
• [12:00] Chilli Cao: its very easy to reproduce
• [12:00] Chilli Cao: it doesnt have to be a fast click as i saw
• [12:00] Chilli Cao: a floater per click
• [12:02] Which Linden: Yes, clearly the pie menu violates the guarantee
• [12:02] Which Linden: I guess it's not a stretch to think that the other UI elements might do so as well
• [12:02] Morgaine Dinova: I don't know how that code is structured, but there are certainly a lot of threads knocking around, and from my work in concurrency that's a recipe for total disaster and mayhem if more than one thread is involved in any of this current issue. Is there more than one involved here?
• [12:03] Chilli Cao: a second click should bring forward the floater thats already opened like when its set to "buy object" when clicked
• [12:03] Which Linden: Morgaine: I think the UI is still single-threaded
• [12:03] Which Linden: Chilli: agreed
• [12:03] Morgaine Dinova: And is that single thread the same thread as the networking?
• [12:03] Latif Khalifa: the grand viewer loop
• [12:03] Latif Khalifa: basically everything is in it
• [12:04] Latif Khalifa: (which is more or less standard "game" design pattern)
• [12:04] Panel 3: Freeze of 0.595888s detected (Mono)
• [12:04] Panel 3: Freeze of 0.584383s detected (LSO)
• [12:04] Panel 3: Freeze of 0.557253s detected (LSO)
• [12:04] Panel 3: Freeze of 0.521248s detected (Mono)
• [12:04] Panel 3: Freeze of 0.544696s detected (Mono)
• [12:04] Panel 3: Freeze of 0.501786s detected (LSO)
• [12:04] Panel 3: Freeze of 0.605135s detected (Mono)
• [12:04] Aimee Trescothick: freezy time
• [12:04] Saijanai Kuhn: Which: [4]
• [12:05] Morgaine Dinova: I've turned it off again, but those have been happening throughout
• [12:05] Which Linden: What's that freeze thing about? Is it an LSL script that just checks for hiccups in the LSL engine?
• [12:06] Which Linden: Cool thanks Sai
• [12:06] Latif Khalifa: its not only the lsl engine
• [12:06] Latif Khalifa: wholse sim freezes up
• [12:06] Which Linden: It's the sim?
• [12:06] Which Linden: Bah
• [12:06] Latif Khalifa: since 1.26 or so
• [12:06] Latif Khalifa: i have observed freezes of up to 60s
• [12:06] Morgaine Dinova: Just tries to get max-rate events, and reports any long intervals. Happens in LSO and Mono continually.
• [12:06] Morgaine Dinova: Yeah, as Latif says
• [12:06] Saijanai Kuhn: [5] also from his blog
• [12:07] Latif Khalifa: it's the biggest source of spoiling the experience for users rightnow
• [12:08] Which Linden: Hm, I presume that there's some d00dz working on it
• [12:08] Which Linden: Is there a PJIRA I can investigate?
• [12:08] Latif Khalifa: could be Which did some nasy blocking call to the back end ;)
• [12:08] Latif Khalifa: SVC-4196
• [12:08] Latif Khalifa: there are tons of jiras
• [12:09] Latif Khalifa: that's one of them with the nice graphs
• [12:09] Latif Khalifa: jira helper died on you? ;)
• [12:09] Latif Khalifa: [6]
• [12:09] JIRA-helper: http//jira.secondlife.com/browse/SVC-4196:
• [#SVC-4196] entering sim: or rezzing object causes sim to freeze for up to 30 seconds - everything stops for everybody there
• [12:09] [[User:JIRA-helper: [#SVC-4196|JIRA-helper: [#SVC-4196]]:
• [12:09] Which Linden: dude JIRA helper
• [12:09] Morgaine Dinova: Don't think anyone's investigating it. At least no Linden with an OH knows of anyone investigating it.
• [12:09] Which Linden: lol
• [12:09] Latif Khalifa: ah its just jira being slow to respond
• [12:10] Latif Khalifa: Andrew said they will assign a dev to look into it "soon" ;)
• [12:10] Which Linden: Oh that's right I saw this in a previous OH
• [12:10] Saijanai Kuhn: I'm heading over for this: [7] session on new mesh imports [8]
• [12:10] Which Linden: Awesome folks, we should wrap up here
• [12:10] Latif Khalifa: 30-60s freezes are bad ;)
• [12:10] Which Linden: Latif: agreed
• [12:10] Morgaine Dinova: And we can't get Babbage's verdict on it, as he's been unavailable for 2+ weeks at his OH
• [12:11] Which Linden: Thanks so much for the interesting discussion
• [12:11] Which Linden: I'll see 'yall next week!
• [12:11] Latif Khalifa: take care which
• [12:11] Chilli Cao: thank you too Which!
• [12:11] Morgaine Dinova: Cheers Which, cya next time
• [12:11] Aimee Trescothick: bye Which :)
• [12:11] Identity Euler: see you
• [12:11] Chilli Cao: that'd be interesting saijanai