Chat log from the meeting on 2022-05-17

From OpenSimulator

Jump to: navigation, search
[11:04] Andrew Hellershanks: Hello, everyone.
[11:06] Vincent.Sylvester Let's see, this week we had few commits, mostly small fixes and cleanup of work that was already going on. We did end up finding a small issue in libomv regarding parsing of data which had some unfortunate consequences for things that relied on that, but that's all resolved now, hopefully.
[11:07] Vincent.Sylvester Title hiding and export perms stuff being worked on, still have to test those things and check with viewer support etc
[11:08] Vincent.Sylvester Seems Firestorm has decided on the way forward regarding the inventory checking code and how to handle things there so that's good
[11:08] Vincent.Sylvester Makes sense to not go completely the route of LL and leave a bit of capability for OpenSim to be more modular
[11:09] Vincent.Sylvester Wiki now has articles on these changes and what they might cause for users as well as how grid operators can attempt to resolve the most problematic issues
[11:09] Kayaker Magic: Can you paste a link to that Vincent?
[11:10] Vincent.Sylvester
[11:10] Kayaker Magic: TX!
[11:10] Andrew Hellershanks: Thank you for the summary of the past weeks activity, Vincent.
[11:11] Ubit Umarov: well about that...
[11:12] Ubit Umarov: Title hide is a option viewers have, that we didn't suport. It should work not
[11:13] Ubit Umarov: VIewers also have a option to not play collision sounds
[11:14] Ubit Umarov: they do it the wronf way. They compare any sound UUID, the server tells them to play, with all the list of collision sounds UUIDs
[11:14] Ubit Umarov: ofc opensim UUIDs where diferent, so that did not work
[11:14] Ubit Umarov: Firestorm did add out UUIDs list
[11:14] Ubit Umarov: our...
[11:15] Ubit Umarov: but as i said that is just BAD, to compare with a list of UUIDs is not only slow but also limited
[11:15] Ubit Umarov: in fact that already does not work with llcollisionSound use
[11:16] Vincent.Sylvester Was about to say
[11:16] Ubit Umarov: will you shutup and listen?
[11:16] Ubit Umarov: well going on..
[11:16] Andrew Hellershanks: Hello, humbletim.
[11:16] humbletim metaverse: hi
[11:17] Ubit Umarov: i did add a flag viewers can set, and it that is present, the region will not send the collision sound
[11:17] Ubit Umarov: that is a more complete solution.
[11:17] Ubit Umarov: now it jsut depends on viewers using, it, and ofc seem most will jsut ignore it
[11:18] Ubit Umarov: they other issue that did pop up and took a lot of talk at FS jira is the new inverntory checks, viewers decided to add, and vicent mentioned
[11:19] Ubit Umarov: well that is just simple.. Those rules are not for Opensim. VIewers should at most Warn
[11:19] Ubit Umarov: i mean some of those rules
[11:20] Ubit Umarov: for example viewers do not allow duplicated system folders, like 2 "objects" folder
[11:20] Ubit Umarov: but having that as always been opensim spec
[11:20] Ubit Umarov: IAR loading will by default do such duplications
[11:21] Ubit Umarov: and ofc too late to go change all saved IAR archives, jsut because some rules changes on viewers...
[11:21] Ubit Umarov: another case is that folders have a numeric type, about its prefered use
[11:22] Ubit Umarov: like My inventory is type 8.. etc
[11:22] Ubit Umarov: suitcase type 100..
[11:23] Ubit Umarov: those numbers main use is to select the pretty icons, and easy the inventory folder  hide things
[11:23] Ubit Umarov: they also help viewer to pic a folder to place a new created item, like a body part...
[11:24] Ubit Umarov: alos so far those things had not been that rigid
[11:24] Ubit Umarov: if we have wrong types, we did get those new itens on wrong place etc
[11:25] Ubit Umarov: to make it worse until 2015 we jsut had those types all wrong...
[11:25] Ubit Umarov: so many issues present on IAR fiels and, even on some older grids DBs
[11:26] Ubit Umarov: again viewers MUST be tolerant to that, since it is jsut impossible for all grids to go fix it
[11:26] Ubit Umarov: Vicent did contribute with some php code to clean some DBs
[11:26] Ubit Umarov: and it is on our Wiki
[11:27] Ubit Umarov: but that only a few can actually use
[11:27] Vincent.Sylvester It just tells you what's wrong, doesn't actually make any DB changes, because that be madness
[11:27] Ubit Umarov: and the number "issues" is large
[11:28] Ubit Umarov: Bew from firestorm did understud that, and converted those hard fails into warnings
[11:28] Ubit Umarov: Beq i meant
[11:28] Ubit Umarov: thing Rye from alchemy jsut made alchemy just work
[11:29] Ubit Umarov: in fact it was Rye who did provide libopenmetaverse and then opensim the fixes for inventory folder types, back in 2015
[11:29] Ubit Umarov: so our action on that is.. no action :p
[11:30] Ubit Umarov: on new code, folders are mostly of the correct type.. etc
[11:30] Ubit Umarov: but IAR loading will go on making duplicates, acording options, like -merge
[11:31] Ubit Umarov: only like 3 folders require some care
[11:31] Ubit Umarov: one it the Current outfit, that in fact as nothing to do with inventory
[11:31] Ubit Umarov: it is appearance storage..
[11:32] Vincent.Sylvester Most issues are on old data from before changes and bugfixes, resolving that isn't something that can be done as just a migration or with a script, it likely has to happen on an individual basis to make sure you aren't bricking peoples inventory completely
[11:32] Ubit Umarov: we mostly only can change things using appearance UI
[11:33] Ubit Umarov: but LL decided to store app on that folder and make it visible, giveing users a another way to break own appearance :p
[11:33] Ubit Umarov: since it is part of appearance, etc,  viewers may be more sensitive about it...
[11:34] Ubit Umarov: it is also nice viewers have no problems finiding "My inventory"
[11:34] Kayaker Magic: Vincent, I'm told that deleting a parent folder can leave child folders orphaned, is that true?
[11:34] Ubit Umarov: rest.. well was working fine..
[11:34] Ubit Umarov: the state of the tree is the only valid thing viewers should actually do
[11:35] Ubit Umarov: parent/child relations is the realy needed thing to keep in good state
[11:35] Vincent.Sylvester Inventory is a cascading structure starting with My Inventory at the top, everything sorted underneath, if you delete a folder any items or folders in there will have no more parent so will fall out of the inventory
[11:35] Ubit Umarov: also if links do point to anything
[11:36] Ubit Umarov: a crash,  bug, etc can affect that ofc
[11:36] Ubit Umarov: and that is what viewers should verifiy
[11:37] Ubit Umarov: well things where working without the new "validation"  firestorm and others jsut did merge from ll
[11:37] Vincent.Sylvester If that causes issues I'll add it to the check script, but so far I haven't seen that be part of the checks, if parent isn't found the stuff seems to just be not part of inventory
[11:38] Ubit Umarov: as i said Beq already converted that into warnings.. so all as it was now
[11:38] Ubit Umarov: well on the next beta :)
[11:38] Ubit Umarov: as i said, seen no issues with alchemy
[11:39] Ubit Umarov: guess rye just did not blind merge that code since it is a area he knows well both on opensim and sl
[11:39] Vincent.Sylvester FS made a floater that tells users the issues and how to get them fixed up with links to the relevant pages so they fix things or have grid operators fix things for them. Most of this is simple SQL queries to fix stuff so just a matter of checking to see what's wrong and running the appropriate query
[11:40] Vincent.Sylvester There is one thing that I been told that might be problematic in different way, viewers matching the folder names to translations rather than doing that by folder type, so if at one point folders were named wrong they might not translate properly
[11:40] Vincent.Sylvester Although I don't know if that's true
[11:40] Andrew Hellershanks: I would expect If a parent is deleted it should be moved to the Trash folder and once Trash is emptied the parent and all items and folders under it should be removed from the avatars inventory. That would avoid leaving entries in the database for folders no longer reachable due to a deleted parent.
[11:41] Vincent.Sylvester That requires additional code on our end to check though
[11:41] Ubit Umarov: Duhh problem is not normal operations
[11:41] Ubit Umarov: it is crashs...   data lost etc etc
[11:41] Ubit Umarov: bugs..
[11:42] Ubit Umarov: inventories like our disks can get damaged
[11:42] Ubit Umarov: seems ll did this because a viwer some did
[11:42] Ubit Umarov: that did a mess of inventories
[11:43] Vincent.Sylvester Like when I tried Kirsten viewer and broke my inventory....
[11:43] Ubit Umarov: but no idea.. they just did, and it is all fine.. FOR SL :p
[11:44] Vincent.Sylvester I found the collision sound thing really weird at first, but it does make sense thinking back
[11:44] Vincent.Sylvester The viewer just gets the sound, doesn't know what it is for
[11:44] Vincent.Sylvester Sending that extra data is a lot of bits
[11:44] Vincent.Sylvester So just not there, otherwise viewer could just block the "type"
[11:45] Vincent.Sylvester Now serverside has to do that instead
[11:45] Ubit Umarov: well i made the new protocol to mute sounds, told a few viewer devs, even sent emails to our mail lists
[11:45] Ubit Umarov: guess 100% ignored..  but lets see :)
[11:45] Vincent.Sylvester It's the best option and frankly more reasonable than using a uuid list when you can set custom sounds
[11:45] Vincent.Sylvester Surprised that is how LL did it, like did they forget about llcollisionsound?
[11:46] Andrew Hellershanks: :)
[11:46] Ubit Umarov: who made that based on uuid list forgot llcollisionsound()
[11:46] Vincent.Sylvester What's next regions exchanging uuid lists of collision sounds with the viewer to make sure they can be blocked
[11:46] Ubit Umarov: you do know 99% of viewer devs just never used or will use SL  right?
[11:47] Ubit Umarov: ll devs i mean
[11:47] Ubit Umarov: tpvs devs are other race.. they love it
[11:47] Ubit Umarov: love  and use sl i mean
[11:47] Vincent.Sylvester I get the same feeling with the mutelist thing in general as to why that is a serverside thing, seems counterintuitive, but eh
[11:47] Ubit Umarov: ll devs are hired ppl
[11:48] Vincent.Sylvester Well after the viewer object cache bug I have lost all faith there
[11:48] Ubit Umarov: fear some just consider it like a cheap game, adding game ideas
[11:48] Ubit Umarov: like right inventory with fixed roles
[11:48] Ubit Umarov: right??  i meant rigid :)
[11:49] Ubit Umarov: ah hand black moons that so light ocean also :p
[11:49] Vincent.Sylvester I'm with you on that, modularity isn't a bad thing
[11:49] Ubit Umarov: seems some ll devs never did comeout at night and looked to the moon
[11:50] Ubit Umarov: and think it if black near horizon
[11:50] Ubit Umarov: but well that is another issue :)
[11:51] Vincent.Sylvester Just odd choices, but that's the norm
[11:51] Ubit Umarov: you do know we may jsut be on borrowed time with sl based viewers
[11:51] Ubit Umarov: they wll stop working on opensim soon or later
[11:52] Ubit Umarov: with the help of ppl like Beq i was able to extend it, adding BOM; EEP; etc..
[11:52] Ubit Umarov: but this will end one day..
[11:52] Ubit Umarov: and rye... and a few others..
[11:53] Vincent.Sylvester Suppose most of it being open source one can always find ways though
[11:53] Andrew Hellershanks: Let's hope that at least one of the OpenSim specific viewers becomes available before that happens, Ubit.
[11:53] Ubit Umarov: who cares it is opensource when takes years to read the code
[11:53] Ubit Umarov: a lot more to actually understand
[11:53] Ubit Umarov: and open source does not mean free to use
[11:54] Andrew Hellershanks: That is one of the big problems with the size of the code base.
[11:54] Vincent.Sylvester Removing the bloat from viewers is going to be the biggest task going forward, because there will be no way around that eventually if there is any hope to get performance out of them
[11:55] Ubit Umarov: originaly LS was created with the ilusion of a open standards etc
[11:55] Ubit Umarov: in fact it was, and to some extent still is
[11:55] Ubit Umarov: what bloat?
[11:55] Ubit Umarov: viewers just have normal software aging
[11:55] Ubit Umarov: opensim code is a LOT worse than viewers code
[11:56] Ubit Umarov: and a lot smaller
[11:56] Ubit Umarov: even so..
[11:56] Vincent.Sylvester I tried setting up a compile of FS before and eventually just had enough from all the stuff it needs to even attempt a compile
[11:56] Vincent.Sylvester Layers add failure points
[11:57] Vincent.Sylvester Like you said, KISS, keeping things simple often means performance and reliability
[11:57] Ubit Umarov: c++ was never a easy thing to do :)
[11:57] Vincent.Sylvester True
[11:57] Ubit Umarov: and a lot worse then like viewer it is cross platform
[11:58] Vincent.Sylvester Between cmake, boost and outdated python I just cringe, perhaps being used to the simplicity of OpenSim compiles
[11:58] Ubit Umarov: but well here we are..  things kinda working.. today :p
[11:59] Vincent.Sylvester Like driving a car that's on fire being held together by speed tape and bubble gum
[11:59] Ubit Umarov: im with the FS beta with thos full rigid validation rules, and had no issues with my inv
[11:59] Andrew Hellershanks: Compiling a viewer has always been a bit of a pain. The first time I tried to do it I gave up. It was a year or more before I tried again and eventually succeeded.
[11:59] Ubit Umarov: ofc never did load a IAR for this avatar...
[12:00] Ubit Umarov: well guess we did cover most code changes since last meeting
[12:00] Ubit Umarov: in fact almost none :)
[12:00] Vincent.Sylvester I probably went at that from the wrong angle trying to automate it into a jenkins node plus trying to figure out how to get it to compile quickly, because waiting half an hour to see if a single change works is just painful
[12:00] Andrew Hellershanks: Ubit, :)
[12:00] Ubit Umarov: -fixed a bub on json serialization
[12:00] Ubit Umarov: also
[12:01] Ubit Umarov: -fixed a bug on json serialization
[12:02] Ubit Umarov: -changed again the exportsupported option internal storage, on some places it needs to be string, not bool
[12:02] Ubit Umarov: -added some try/cacth to try guard agains null ref of mantis 8998
[12:03] Ubit Umarov: - and the already talked collision mutes and title hide
[12:03] Ubit Umarov: viewers also have a gestures sounds thing
[12:04] Ubit Umarov: that is not done by a list compare, bc there is none :p
[12:04] Ubit Umarov: even so, sure sure it is correct..  fear not
[12:04] Ubit Umarov: fear it jsut mutes any sounds with the avatar as origin
[12:05] Ubit Umarov: but thats 100% viewer side thing....
[12:05] Ubit Umarov: just had not seen it till i seen the collisions one :)
[12:05] Vincent.Sylvester Aren't gesture sounds "dropped" and don't move with the avatar?
[12:05] Ubit Umarov: yeah
[12:06] Ubit Umarov: think viewers just kill any sounds like that
[12:06] Ubit Umarov: bc there is no tag telling "im a gesture sound"
[12:06] Ubit Umarov: but well viewer thing
[12:07] Vincent.Sylvester All comes down to not wanting to send a few bits with each sound to specify what type it is, makes sense in 2007, but now...
[12:07] Vincent.Sylvester Precious bandwidth heh
[12:07] Ubit Umarov: well question is should there be any mute at all?
[12:08] Ubit Umarov: you don't spend time creating a immersion enviroment, then go add options per user to disable parts of it
[12:08] Ubit Umarov: that is a BIG mistake ll did since ever
[12:09] Ubit Umarov: like the basic always midday setting
[12:09] Vincent.Sylvester I get the point of doing it for people that have disabilities that may cause them issues, but if it was about that the execution is lacking severely
[12:09] Ubit Umarov: region creator had all the work making day and night composition
[12:09] Andrew Hellershanks: The ability to mute has its uses.
[12:10] Ubit Umarov: then users go there with fixed midday
[12:10] Ubit Umarov: nonsense
[12:10] Vincent.Sylvester I just turn my volume down
[12:10] Ubit Umarov: ofc all has uses
[12:10] Ubit Umarov: but are we building a 3d metaverse or what?
[12:11] Ubit Umarov: well options to make all ladies naked are cool also
[12:11] Ubit Umarov: should we really have them?
[12:11] Ubit Umarov: then on the other hand options like display LookAt ???
[12:12] Ubit Umarov: forgetting that lookat IS NEEDED
[12:12] Ubit Umarov: lol
[12:12] Ubit Umarov: head does not move without
[12:13] Ubit Umarov: guns aim go not work
[12:13] Vincent.Sylvester I guess the look at thing was to see who was being weird, but that's pointless as the camera doesn't work that way
[12:13] Ubit Umarov: avatar rotation without head is low resolution rotation
[12:14] Ubit Umarov: the option to display it was the worng option
[12:14] Ubit Umarov: correct action was to just remove showing it
[12:14] Ubit Umarov: not not sending it
[12:15] Ubit Umarov: but well minor things
[12:15] Vincent.Sylvester A lot of these ideas seem good on paper, but need to be tested to see what consequences they have, QA never been LL strong point
[12:15] Ubit Umarov: so. what news do oyu bring about opensim?
[12:15] Vincent.Sylvester lol
[12:16] Ubit Umarov: welcome humbletim
[12:16] Andrew Hellershanks: Ubit, What else is there other than the summaries that you and Vincent provided?
[12:16] Ubit Umarov: well that is what i asked :P
[12:17] Andrew Hellershanks: We are quarter past the hour. Does anyone have a question they have been waiting to ask during todays meeting?
[12:17] Ubit Umarov: we know kayker is breaking sims with raycast :p
[12:18] Andrew Hellershanks: :)
[12:18] Andrew Hellershanks: How is that going, Kayaker?
[12:18] Ubit Umarov: well terrain raycast are also a reason why ODE needs a huge stack setting on linux
[12:19] Ubit Umarov: ( ulimit -s )
[12:19] Kayaker Magic: I was traveling last weekend, camping with my 'bros in RL this weekend. Will have time to narrow it down this week.
[12:19] Ubit Umarov: it is very heavy code currently
[12:19] Ubit Umarov: wel since there is ODE lib
[12:19] Jamie.Jordan           *•.¸'*•.¸ ★ ¸.•*´¸.•*
[12:19] Jamie.Jordan  .•*★¨`•  Aρρlɑuѕɛ  •¨`★*•.
[12:19] Jamie.Jordan      ¸.•*`¸.•*´  ★  `*•.¸`*•.¸
[12:19] Jamie.Jordan SORRY
[12:19] Andrew Hellershanks: Kayaker, ok, sounds good. I wouldn't mind getting away but that's not going to happen any time soon.
[12:20] Vincent.Sylvester I did look at that and there is meant to be something that stops checking for things when it encounters terrain
[12:20] Vincent.Sylvester Not sure how but that might be causing problems
[12:21] Vincent.Sylvester Haven't had much more time to read that, probably just have to add tons of debug output and break it
[12:21] Ubit Umarov: no it is mostly plain bad code on ODE
[12:21] Ubit Umarov: i have on TODO, to change it for years :)
[12:22] Ubit Umarov: i may do it.. one day ;)
[12:22] Ubit Umarov: terrain on ODE is very very bad
[12:22] Ubit Umarov: remember how old ODE module lets us fall from terrain?
[12:23] Vincent.Sylvester You can still do that
[12:23] Ubit Umarov: well that shows how bad is that code
[12:23] Ubit Umarov: i did improve that for ubode..
[12:23] Vincent.Sylvester Set prim to phantom via the volume detect then set it physical and sit on it
[12:24] Vincent.Sylvester Results are hilarious
[12:24] Ubit Umarov: guess not issue on that
[12:24] Ubit Umarov: you just sit like on bullet
[12:24] Ubit Umarov: bc the prim stays phanton
[12:24] Ubit Umarov: you need to clear the phantom
[12:25] Ubit Umarov: thats a thing on volume detect
[12:25] Vincent.Sylvester Yeah but eventually the prim goes out of bounds and gets deleted so you are forced to stand up
[12:25] Ubit Umarov: it needs to set the prim globla phanton, but cant take it off, because reasons
[12:25] Vincent.Sylvester OpenSim without some funky bugs just isn't fun
[12:25] Ubit Umarov: well physical should collide with terrain
[12:26] Ubit Umarov: but who tells you to move from a volume detector into a physical prim?
[12:26] Ubit Umarov: duhh  :p
[12:27] Vincent.Sylvester You know I attract bugs like that for some reason
[12:27] Vincent.Sylvester Not like I actively look for them either
[12:27] Vincent.Sylvester Never won the lottery either go figure
[12:27] Ubit Umarov: :)
[12:29] Andrew Hellershanks: :)
[12:30] Selby.Evans must go  -- bye all
[12:30] Andrew Hellershanks: We have been going for an extra half hour. Time to wrap up the meeting for today.
[12:30] humbletim metaverse: hey gotta run to another meeting too -- thanks all and enjoyed listening in today. :)
[12:30] Andrew Hellershanks: Thank you all for coming. See you again next week.
[12:30] Andrew Hellershanks: Bye, Selby.
[12:30] Ubit Umarov: cya selby.Evans
[12:30] Motoko.Karu thanks everyone...
[12:30] Motoko.Karu bye
[12:30] Andrew Hellershanks: humbletim, np. Nice to meet you. Drop in again some time.
Personal tools
About This Wiki