Content Creation/Scripting User Group/Transcripts/2011 05 09

From Second Life Wiki
Jump to navigation Jump to search

List of Speakers

flexi campfire
Latif Khalifa
Liisa Runo
Stickman Ingmann
Techwolf Lupindo
Void Singer

Transcript

[09:03] Stickman Ingmann: Yay, Kelly's here!

[09:03] tehKellz (kelly.linden): Sorry for being late.

[09:03] Void Singer cancels the flogging

[09:03] tehKellz (kelly.linden): How is everyone doing?

[09:04] Stickman Ingmann: Pretty good.

[09:04] draconis.neurocam (draconis.neurocam): good, yourself?

[09:04] tehKellz (kelly.linden): doing all right.

[09:04] Void Singer: I've been up for ~ 36hrs, but I'm still conscious

[09:04] tehKellz (kelly.linden): o_O

[09:04] Liisa Runo is good. Bought 18 bags of seeds today, my RL garden will be full of pretty flowers and tasty herbs this summer

[09:04] draconis.neurocam (draconis.neurocam): jolt cola?

[09:04] tehKellz (kelly.linden): nice Liisa.

[09:05] Void Singer: cafienne and nictone ftw

[09:05] Stickman Ingmann: You can grow caffeine?

[09:05] tehKellz (kelly.linden): I dunno. I kinda like sleep.

[09:05] Stickman Ingmann: Oh, I'm mixing topics.

[09:05] Void Singer: Stick, sure

[09:05] Techwolf Lupindo: Nothing like a good mix there.

[09:05] Techwolf Lupindo: :-)

[09:06] Stickman Ingmann: The guarana berry is naturally rich in caffeine, but requires a very temperate climate. Oh look, scripting usergroup.

[09:06] ??? Ca?? (nad.warrior): LINDENZ

[09:07] tehKellz (kelly.linden): :)

[09:07] ??? Ca?? (nad.warrior): you are a disgrace to your race

[09:07] Stickman Ingmann: I remember this guy. He was here the last time I was.

[09:07] Stickman Ingmann: So what's on the agenda, Kelly? I had a few questions, but don't want to go before you do.

[09:07] ??? Ca?? (nad.warrior): Nooo one forget the imperator

[09:08] tehKellz (kelly.linden): So I sent out an email to a few lists on Friday about the changes currently on Aditi in the mono upgrade branch.

[09:08] ??? Ca?? (nad.warrior): bitch you freeze me again and i'll break that ass

[09:08] ??? Ca?? (nad.warrior): ok bye =]

[09:08] Stickman Ingmann: Do Lindens have a "make life hell" option on the pie menu? That'd be neat.

[09:09] tehKellz (kelly.linden): I dunno where it all went in viewer 2.

[09:09] tehKellz (kelly.linden): or web profiles

[09:09] tehKellz (kelly.linden): anyway

[09:09] Haravikk (haravikk.mistral): Maybe an option to accidentally leave their credit card details on a train somewhere?

[09:09] tehKellz (kelly.linden): haha

[09:09] Stickman Ingmann: Has Latif come back with benchmark results for the new mono patch yet?

[09:09] Void Singer: I head nothing and my logs will show as musch

[09:10] Void Singer: heard*

[09:10] tehKellz (kelly.linden): Not that he has shared with me.

[09:10] tehKellz (kelly.linden): I haven't had a chance to look at jira yet this morning for new bugs.

[09:11] Haravikk (haravikk.mistral): Is this Mono update related to the version with llRegionSayTo()? I only found out about that yesterday, seemed to appear very quickly and quietly, I only missed one meeting I think =)

[09:11] Void Singer: I adde one to the agenda hope you don't mind

[09:11] tehKellz (kelly.linden): Nope. llRegionSayTo wasn't talked about a lot at this meeting

[09:11] Stickman Ingmann: Where's the agenda?

[09:11] Qie (qie.niangao): ( ! There's an agenda? )

[09:11] tehKellz (kelly.linden): I don't even know where the agenda is anymore.

[09:12] Stickman Ingmann: Ama Omega's profile has an old copy.

[09:12] Void Singer: lol I tacked it on the old page

[09:12] Stickman Ingmann: The wiki's currently being broken.

[09:12] tehKellz (kelly.linden): Right. let me find it again

[09:12] Void Singer: scr-66

[09:12] flexi campfire: http://jira.secondlife.com/browse/SCR-66

[#SCR-66] llRegionSayTo() Fails to send messages to attachments on sitting avatars

[09:12] Stickman Ingmann: Ah, I can see how that bug would crop up. Fun.

[09:12] Void Singer: llRegionSayTo fails when targeing a sittin av

[09:13] tehKellz (kelly.linden): oh. fun.

[09:13] Stickman Ingmann: While we're on Jira, https://jira.secondlife.com/browse/SVC-4011 should probably be marked as "fixed."

[09:13] flexi campfire: [#SVC-4011] Please could he have a function llSetLinkGlow & llSetGlow (or even llSetLinkAlphaAndGlow & llSetAlphaAndGlow)

[09:13] tehKellz (kelly.linden): I will pass that on.

[09:13] Haravikk (haravikk.mistral): Does it fail completely, isn't it just the attachments that don't get reached?

[09:13] Abby Normal (frida.rasmuson): I did a little more testing. it does NOT send the message to the sat upon prim

[09:13] Void Singer: Hav for all intents and purpose, complete failure

[09:14] Void Singer: I didn't double check 0 channel

[09:14] Abby Normal (frida.rasmuson): 0 dosn't go to attachments

[09:14] Haravikk (haravikk.mistral): I know it's a major issue but I think the av is still reachable on channel zero, just not the attachments (on any channel)

[09:15] Haravikk (haravikk.mistral): It's like the avatar is being treated as a child-prim while sitting instead of using the correct avatar behaviour

[09:15] tehKellz (kelly.linden): All right. Shouldn't be that difficult to fix.

[09:15] Abby Normal (frida.rasmuson): Woot...

[09:15] tehKellz (kelly.linden): yeah. because at the code level a sitting agent is a child prim.....

[09:15] Void Singer: cool, I was hoping it was simple... does that mean there should be no bleed to the sat upon linkset?

[09:16] tehKellz (kelly.linden): yeah, there shouldn't be.

[09:16] Void Singer: cool

[09:16] tehKellz (kelly.linden): it is just a matter of adjusting the parent logic for attachments.

[09:16] Abby Normal (frida.rasmuson): Void the sat upon object does not hear the message at the present time

[09:17] Void Singer: Frida I just wanted to make sure that didn't change is why I noted it

[09:17] tehKellz (kelly.linden): I bet if you sent llRegionSayTo to the vehicle all the agents attachments would get it

[09:17] Haravikk (haravikk.mistral): One thing I forgot to check was wether all messages are being directed to the viewer (including non-zero ones) while the avatar is sitting, as that could make it even more serious

[09:17] tehKellz (kelly.linden): only channel 0 (and script_debug) are ever sent to the viewer

[09:17] Abby Normal (frida.rasmuson): can viewers even see non 0 channels?

[09:17] tehKellz (kelly.linden): nope

[09:18] Abby Normal (frida.rasmuson): I didn't think so

[09:18] Void Singer: not without help from a script

[09:18] Haravikk (haravikk.mistral): Not normally anyhow, but if the sim thinks the avatar is a prim...?

[09:18] Abby Normal (frida.rasmuson): nods nods

[09:18] Stickman Ingmann: Kelly, would that be desired behavior? Avatar attachments getting it when the message is sent to something they're sitting on?

[09:18] Latif Khalifa: viewers only "hear" two chanles, 0 and debug

[09:19] tehKellz (kelly.linden): Stickman I don't think so.

[09:19] Haravikk (haravikk.mistral): Would avatar attachments get the vehicle's messages? They're not really prims in a normal sense, and not strictly children of the avatar are, they just happen to be on it

[09:19] Void Singer: that could be a minor security issue also

[09:19] Latif Khalifa: nothing much uses that function yet

[09:20] Stickman Ingmann: Can still play with the rules, yeah.

[09:20] tehKellz (kelly.linden): Oh. Hey Latif.

[09:20] Haravikk (haravikk.mistral): I think a message directed at an object shouldn't go to an avatar sat on it, should be up to the object if it wants to pass it on

[09:20] Latif Khalifa: hey Kelly

[09:21] tehKellz (kelly.linden): Someone was asking for your results on testing the new mono stuff.

[09:21] tehKellz (kelly.linden): Haravikk: I agree.

[09:21] Stickman Ingmann: That was me trying to harass you Latif. You didn't respond, I thought you had died.

[09:21] Latif Khalifa: i really liked the one which was setup at 3 script rezzes per frame. seemed optimal to me

[09:22] tehKellz (kelly.linden): That is what we are going with (and is deployed to all mono 2 upgrade regions on aditi now)

[09:22] Haravikk (haravikk.mistral): I'm wondering what should be done about the llTargetSay() issue in light of llRegionSayTo(), since it does the same thing (albeit in a buggy way since llTargetSay() was only intended to message a single prim ;)? I forget the JIRA number

[09:22] Haravikk (haravikk.mistral): Mark it as fixed?

[09:22] Latif Khalifa: it totally eliminates sim freeze on avatar entry or rez with large number of scripts.

[09:23] tehKellz (kelly.linden): Haravikk: yeah that is probably a good idea.

[09:23] Abby Normal (frida.rasmuson): As an aside, finding this meeting through the office hours page gave me a time of 11-12 and a location on Aditi

[09:23] Stickman Ingmann: https://jira.secondlife.com/browse/SVC-4011 is in the same boat. We have the update to llSetPrimitiveParams that does those now.

[09:23] flexi campfire: [#SVC-4011] Please could he have a function llSetLinkGlow & llSetGlow (or even llSetLinkAlphaAndGlow & llSetAlphaAndGlow)

[09:23] tehKellz (kelly.linden): SVC-92

[09:24] tehKellz (kelly.linden): http://jira.secondlife.com/browse/SVC-92

[09:24] Haravikk (haravikk.mistral): llTargetSay() is SCR-2, I don't seem to have the option to close it as fixed (or close it at all)

[09:24] flexi campfire: http://jira.secondlife.com/browse/SCR-2

[#SCR-2] llTargetSay() - region-wide direct communication

[09:24] tehKellz (kelly.linden): k, closed it.

[09:25] tehKellz (kelly.linden): looks like the campfire can't handle redirects

[09:25] Haravikk (haravikk.mistral): Thanks!

[09:26] Void Singer: speaking of redirect, I had a SVC jira moved to EXP... rather than cloned SVC-25555 it's now hidden to regular users

[09:26] tehKellz (kelly.linden): Hunh. How bizarre. That SVC-4011 doesn't have the 'flow' options I think it should.

[09:27] Stickman Ingmann: I had a question about script memory and LSL/Mono, Kelly. I'm pretty sure I'm right, I just want to double check before I go trying to defend some of my choices to the public. I wanted to confirm two situations.

[09:27] tehKellz (kelly.linden): Void: is that number correct? I can't find it at all.

[09:27] tehKellz (kelly.linden): Stick: go ahead

[09:28] Void Singer: sorry that should have been VWR not SVC

[09:28] Stickman Ingmann: First, the case of 10 unique scripts, that are 4k each in both LSL and Mono -- assuming absolutes for simplicity's sake. LSL reports 160KB used. This is correct, as LSL uses 16KB per script, regardless of how much it actually uses, right? Mono reports 1024KB -- 64k maximum per script. But it's really using 4*16, or 64KB. Is that the case?

[09:28] Stickman Ingmann: Heh. My math is broken.

[09:29] Stickman Ingmann: 4*10, 40KB.

[09:29] tehKellz (kelly.linden): Correct.

[09:30] Techwolf Lupindo: 640KB should be enought.

[09:30] Stickman Ingmann: Second is the case of 10 scripts with the same UUID. We'll assume 3k bytecode and 1k used memory. LSL reports 160KB, which is accurate. Mono reports 1024 -- but it's actually 3+(1*10). Is that the case?

[09:30] tehKellz (kelly.linden): Void: I dunno what 'flow' the exp team is using. :-/ the issue is still open.

[09:30] Stickman Ingmann: 10 identical scripts, rather. Babbage says that the same UUID is required for shared bytecode.

[09:30] tehKellz (kelly.linden): the same asset id is required for shared bytecode.

[09:30] Stickman Ingmann: Recompiling an identical script results in a new UUID, so no shared bytecode.

[09:31] tehKellz (kelly.linden): right

[09:31] Latif Khalifa: which you can get if you drop same mono script from the inventory without recompiling.

[09:31] tehKellz (kelly.linden): right again.

[09:31] Stickman Ingmann: Alright. Wanted to make sure I had those right.

[09:31] Qie (qie.niangao): Back to memory usage: I still have a bit of confusion... there's some notion of the Mono script having "reserved" 64KB, right? regardless of how much it's using?

[09:31] tehKellz (kelly.linden): Correct Qie.

[09:32] tehKellz (kelly.linden): You can use llGetUsedMemory on the new mono branch (or the profiling functions there) to see how much is actually used.

[09:32] Stickman Ingmann: So if you have 1000 Mono scripts, does the simulator reserve 64MBs of memory for them that nothing else can use?

[09:32] Qie (qie.niangao): so... I'm not sure what happens to "reserved" but "unused" memory... is it still mapped to virtual? or... well, you see the confusion.

[09:32] tehKellz (kelly.linden): No, it is not mapped or used.

[09:32] Stickman Ingmann nods.

[09:32] Qie (qie.niangao): ah, okay.

[09:33] Latif Khalifa: "reserve" is more of an accounting notion for script limits

[09:33] Qie (qie.niangao): thanks.

[09:33] tehKellz (kelly.linden): The memory limits on mono scripts are an artificial construct created by us.

[09:33] Latif Khalifa: sort of like "free prims on a parcel"

[09:33] Stickman Ingmann: I have not heard about llGetUsedMemory(), Kelly. Is this on Aditi?

[09:33] tehKellz (kelly.linden): We track the memory used by scripts by walking the accessible memory of the script.

[09:34] Techwolf Lupindo: isn't that an expensive process?

[09:34] tehKellz (kelly.linden): Yes it is on aditi. Along with llScriptProfiler, PROFILE_SCRIPT_MEMORY and llGetSPMaxUsedMemory

[09:34] tehKellz (kelly.linden): Yes. We don't do it that often.

[09:35] Latif Khalifa: Stickman, there are a couple of nice new profiling functions. You can basically tell script. OK: start monitoring now, ok done, tell me what happened with memory during that time.

[09:35] Liisa Runo: http://wiki.secondlife.com/wiki/LlGetUsedMemory and http://wiki.secondlife.com/wiki/LlScriptProfiler

[09:35] tehKellz (kelly.linden): Unless you turn on profiling in the new code. Then we do it all the time.

[09:35] Stickman Ingmann: I am going to need to play with these.

[09:35] tehKellz (kelly.linden): And it is really slow.

[09:35] Stickman Ingmann: Thank you for informing me.

[09:35] tehKellz (kelly.linden): (10-100x slower depending on the size of your script)

[09:35] Latif Khalifa: the nice thing it ca be turned on and off on runtime

[09:36] Haravikk (haravikk.mistral): How is it slowed, by counting the profiling as script-instructions?

[09:36] tehKellz (kelly.linden): It counts the memory more closely to record a 'max used' value.

[09:37] tehKellz (kelly.linden): Normally we keep a rough running total of allocations only (since mono uses a garbage collector it isn't easy to track deallocations) and when our approximate value gets past 64k we do an accurate count.

[09:37] tehKellz (kelly.linden): When profiling is on we don't use the approximate at all.

[09:38] tehKellz (kelly.linden): When you call llGetUsedMemory we calculate the memory right then.

[09:38] Abby Normal (frida.rasmuson): well the thing is people believe , having been told so, that ALL mono scripts always used 64KB and froth at the mouth about 'wasting "my" regions resources'.

[09:38] tehKellz (kelly.linden): so it isn't the fastest library call but you don't need profiling enabled to use it.

[09:38] tehKellz (kelly.linden): They have been told incorrectly.

[09:39] Stickman Ingmann: That's what I'm going to be fighting against, Frida. It's a common misconception.

[09:39] Abby Normal (frida.rasmuson): It seems to be 'institutional knowledge' now

[09:39] Haravikk (haravikk.mistral): Mono has never allocated the full 64kb to a script right away I don't think?

[09:39] tehKellz (kelly.linden): All LSL2 scripts use 16k always. Mono scripts use what they need.

[09:39] Void Singer: that myth is a result of the script info window mostly

[09:39] Abby Normal (frida.rasmuson): We're all fighting that...

[09:40] Liisa Runo: (some people still believe that 20meter cube cause more lag than 10meter cube)

[09:40] Qie (qie.niangao): (bearing in mind that it doesn't take much of a string to use up 4K in Mono.)

[09:40] Abby Normal (frida.rasmuson): NO That I understand (using up memory)

[09:40] Haravikk (haravikk.mistral): I'm still curious about the speed change with profiling, does the accurate account occur every operation, or every operation that could change memory usage

[09:40] tehKellz (kelly.linden): This has been a point of debate, of course. When we limit script memory we were aiming for reducing the surprise and content breakage that dynamic memory could lead to. As such the limits would be based on 'reserved' memory so that no script could change size and force things to break.

[09:41] Stickman Ingmann: I have a script that uses 11KB in Mono. Requires very limit working memory, so it's mostly bytecode. I need 10 copies of it in my new avatar. That's 160KB actual for LSL. Or considerably less in Mono -- but Mono would report 640KB used with the old functions.

[09:41] tehKellz (kelly.linden): Haravikk: neither. We check many times a basic functions 'checkMemory()' that will set the stack-heap if needed.

[09:42] tehKellz (kelly.linden): Stickman: for serialization (crossing region borders) there is no benefit to the shared bytcode)

[09:42] Liisa Runo: SVC-3209

[09:42] flexi campfire: http://jira.secondlife.com/browse/SVC-3209

[#SVC-3209] llLinkTargetOmega() and/or PRIM_OMEGA flag for PrimParam functions

[09:42] Stickman Ingmann: Good to know.

[09:42] tehKellz (kelly.linden): Haravikk: that function when profiling is disabled does some very lazy accounting.

[09:42] tehKellz (kelly.linden): Especially if you aren't allocating a whole lot.

[09:42] tehKellz (kelly.linden): With profiling enabled every time we hit that we do an accurate account.

[09:43] tehKellz (kelly.linden): It isn't every instruction, but it is way more frequent then the lazy method.

[09:43] Haravikk (haravikk.mistral): Right I think get I it, just wasn't quite sure when the profiling is actually occurring to cause the slow down. Say a script does nothing but math with no lists or temporary variables, would the profiling add any major overhead in that case?

[09:44] tehKellz (kelly.linden): yes.

[09:44] tehKellz (kelly.linden): That is actually when it would do the most.

[09:44] tehKellz (kelly.linden): Because the lazy accounting would do an accurate count very rarely if you aren't allocating memory

[09:44] tehKellz (kelly.linden): But the profiling will do it frequently

[09:45] tehKellz (kelly.linden): And the performance hit is going to relate to how much memory you have allocated already too. So if you just do math but also don't have any lists or strings sitting around you would probably be ok.

[09:45] tehKellz (kelly.linden): but if you have a large script and do a bunch of math it is going to be very, very slow.

[09:46] tehKellz (kelly.linden): that is pretty much the worst case scenario for the profiling.

[09:46] Haravikk (haravikk.mistral): k! Thanks for explaining =)

[09:46] !< Sofa: Click the seat to change your animation or position.

[09:47] tehKellz (kelly.linden): Ok 15 mins, did I miss any questions or topics?

[09:47] Liisa Runo: SVC-3209 for my pretty retro bicycle. Nothing to discuss about, we just want it

[09:47] draconis.neurocam (draconis.neurocam): i agree with liisa, my staff could be one script if we had it

[09:47] tehKellz (kelly.linden): k

[09:48] Void Singer: looking for early feedback on an estate level switch to make temp rezzing more friendly for full combat regions

[09:48] Haravikk (haravikk.mistral): Would that throttle the rate at which temp-on-rez objects can be created?

[09:48] Void Singer: dethrottle for those regions actually

[09:49] Abby Normal (frida.rasmuson): No what currently happens is that when physics falls temp-rez gets cueued.

[09:49] Void Singer also wants Liisa's function

[09:49] Abby Normal (frida.rasmuson): then the all get released at once.

[09:49] tehKellz (kelly.linden): what would you propose we do to make temp rezzing more friendly?

[09:49] Abby Normal (frida.rasmuson): not queue them.. In the combat regions

[09:50] draconis.neurocam (draconis.neurocam): could be a simconsole option

[09:50] Void Singer: reduce/ remove queing?

[09:50] Abby Normal (frida.rasmuson): the flushed queue can end up woth 500 odd objects at once. the woest case for the current combat systems in uses is about 240 objects

[09:50] Void Singer: I'm going to pitch it to Oskar as well, but I wanted to know your thoughts Kelly

[09:51] tehKellz (kelly.linden): so when physics takes a dive would we .... just drop the rez requests on the floor?

[09:51] Haravikk (haravikk.mistral): Hmm, would it make sense to have a per-user throttle to coincide with it though? Seems like if the physics are slowing enough to queue temp-on-rez then that's the real problem, so something to help out should really be added as well

[09:51] tehKellz (kelly.linden): This is the first I've heard of the issue.

[09:51] Void Singer: the caching is seeming to queue a lot more that would have actually been instanciated in the same time

[09:51] Abby Normal (frida.rasmuson): 60 people running around a soin will reduce physics to a crawl

[09:52] Latif Khalifa: it's relatively recent change (i always blame falcon). rezzing a lot of objects (like shooting arrows/bullets whatever) gets queued so they don't appear, and then then 100s rez at the same time.

[09:52] tehKellz (kelly.linden): I didn't know we queued temp rezzes at all.

[09:52] Haravikk (haravikk.mistral): True, but I think without some kind of complementary ability to restrict temp-on-rez creation you could end up asking for trouble

[09:52] Latif Khalifa: one of those "optimizations" that makes things worse

[09:52] Void Singer: going to try to put together a test case for accurate figures

[09:52] Abby Normal (frida.rasmuson): Yes......... there is even a JIRA on it

[09:53] Abby Normal (frida.rasmuson): one sec

[09:53] tehKellz (kelly.linden): so what would be the behavior change? Don't queue? Drop the requests (don't rez them)?

[09:53] Latif Khalifa: revert to old behavour. rez them as they come

[09:53] Latif Khalifa: even it it means sim will perform a bit slower

[09:53] Abby Normal (frida.rasmuson): SVC-5927

[09:53] flexi campfire: http://jira.secondlife.com/browse/SVC-5927

[#SVC-5927] Temp on Rezzed objects get queued

[09:53] Void Singer: Haravikk, that's why I was thinking estate level switch... so that region managers could choose whether or not it was a benifit

[09:53] tehKellz (kelly.linden): hm. I think you'll need to bring it up at Andrew's meetings.

[09:53] Latif Khalifa: yeah this is for andrew/simon/falcon

[09:54] Abby Normal (frida.rasmuson): Andrew seems to think that it would affect the other regions on a server.

[09:54] Abby Normal (frida.rasmuson): I don;t see how it can

[09:54] tehKellz (kelly.linden): it isn't that new of an issue apparently - nearly a year old.

[09:54] tehKellz (kelly.linden): oh. I bet we block based on memory used.

[09:54] Stickman Ingmann: Andrew meets tomorrow at noon (SL time) and ... Thursday at 4pm?

[09:54] Latif Khalifa: my sense of "recent" is b0rked

[09:55] Stickman Ingmann: In the Denby region.

[09:55] Abby Normal (frida.rasmuson): and if you will notice people starts screamin immediately

[09:55] Haravikk (haravikk.mistral): It does sound like queuing is a silly solution, as rather than really reducing strain you just seem to end up pushing into bursts of activity instead

[09:55] Haravikk (haravikk.mistral): I can't help but feel there should be a better solution than queuing in general, rather than specifically for temp-on-rez

[09:55] Latif Khalifa: yeah it's pretty crippling for some types of sims

[09:56] Abby Normal (frida.rasmuson): nioce understatment

[09:56] Abby Normal (frida.rasmuson): Oh Kelly, thank you for putting a copy if Sardar on aditi

[09:56] tehKellz (kelly.linden): Probably thank Oskar or Maestro.

[09:56] Abby Normal (frida.rasmuson): we're setting it up so the people can come an test combat in a 'real' environment

[09:56] Haravikk (haravikk.mistral): Oh, gtg, thanks all, see you next week =)

[09:57] tehKellz (kelly.linden): o/

[09:57] tehKellz (kelly.linden): Last minute question from me then:

[09:57] Abby Normal (frida.rasmuson): nods nods

[09:57] tehKellz (kelly.linden): I've been considering ways to get better testing for these scripting server updates after my last one was rolled back twice two weeks in a row

[09:58] tehKellz (kelly.linden): If I got a set of sandboxes up with the update, would that get used more than the aditi tests?

[09:58] tehKellz (kelly.linden): sandboxes on agni.

[09:58] draconis.neurocam (draconis.neurocam): if people knew about them i imagine yes

[09:58] Ima Mechanique (nielarcher): not for me ;-)

[09:58] Abby Normal (frida.rasmuson): yes.... the problem with Aditi is thet it is non trivial to get to.

[09:58] Liisa Runo: i would surely do more testing if i could do it in main grid

[09:58] Void Singer: almost definitely

[09:58] tehKellz (kelly.linden): That might be the trick. Though, no more of a trick than letting em know about what is on aditi.

[09:58] Qie (qie.niangao): personally, I'd sometimes get to the agni sandboxes... I just never get to Aditi, no matter how much I intend to.

[09:59] Ima Mechanique (nielarcher): I spend more time on Aditi working, than agni

[09:59] tehKellz (kelly.linden): k, that seems to mirror my thoughts. I'm still debating between full RC this week and a set of agni sandboxes for the mono upgrade.

[09:59] Latif Khalifa: i think that's a great idea. getting products on aditi is a bit of a hassle. many scripters are eager to test their stuff. having sandboxes on agni would help a lot.

[10:00] Void Singer: could drop one next to the existing sandboxex and drop some warning signs in it

[10:00] Latif Khalifa: i think the biggest problem is finding an effective way to communicate about it.

[10:00] tehKellz (kelly.linden): Yes. Communication is an issue.

[10:00] tehKellz (kelly.linden): Speaking of, I have an important meeting to attend now.

[10:00] tehKellz (kelly.linden): Thanks all for coming out.