(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
[12:18 PM PT] Ubit Umarov: welcome :)
[12:18 PM PT] Ubit Umarov: hmm lets see last week code changes..
[12:19 PM PT] Gavin.Hird @grid.xmir.org:8002: I haded a image to a couple of you and here is the version info the viewer picks up
[12:20 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Yeah was about to say we do send some things on caps for simulator info as well
[12:20 PM PT] Vincent.Sylvester @hg.zetaworlds.com: With script functions you can get some more info out even what memory usage is
[12:20 PM PT] Ubit Umarov: ok added function llIsFriend
[12:21 PM PT] Ubit Umarov: and made several changes on supporting code it needs, updating it a bit to dotnet 6 features
[12:22 PM PT] Ubit Umarov: made a few changes to make region a bit more resilent to joe "small" bandwidth requests :)
[12:22 PM PT] Ubit Umarov: well ther where needed there anyways
[12:23 PM PT] Ubit Umarov: and a few more, just "cosmetic" code changes
[12:23 PM PT] Ubit Umarov: linkset data still has no persistence .. it not stored on dbs or oars or leaves region
[12:24 PM PT] Gavin.Hird @grid.xmir.org:8002: I ahve all your changes running on my grid, and it seems to work, except I have not tested the llIsFriend function
[12:24 PM PT] Gavin.Hird @grid.xmir.org:8002: been busy buying house...
[12:24 PM PT] Ubit Umarov: :)
[12:24 PM PT] Ubit Umarov: wait to the move..
[12:24 PM PT] Gavin.Hird @grid.xmir.org:8002: handover is March 15
[12:24 PM PT] Ubit Umarov: :)
[12:25 PM PT] Vincent.Sylvester @hg.zetaworlds.com: We ought to add a check if the friendslist is empty and return "You ain't got no friends loser" xD
[12:25 PM PT] Gavin.Hird @grid.xmir.org:8002: :-)
[12:25 PM PT] Gavin.Hird @grid.xmir.org:8002: one can always make an alt and friend it
[12:25 PM PT] Ubit Umarov: wel like at sl, you are not your friend!
[12:25 PM PT] Ubit Umarov: :)
[12:26 PM PT] Vincent.Sylvester @hg.zetaworlds.com: SL doesn't believe in self love, no surprises there
[12:26 PM PT] Gavin.Hird @grid.xmir.org:8002: no, but testing the functionality with an alt should work fine
[12:26 PM PT] Vincent.Sylvester @hg.zetaworlds.com: We got a few more LSL functions that are in the works, but I got sick of LL changing their minds and stopped working on them for now
[12:26 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Will pick that up again in a couple weeks I think
[12:27 PM PT] Ubit Umarov: joe you cna change that script to use llOwnerSay and on region change event..
[12:27 PM PT] Ubit Umarov: many regions will allow scritps on attachments to run..
[12:28 PM PT] Ubit Umarov: guess you attach it to your test avatar using fs, it will run when you test with your viewer
[12:29 PM PT] Say Version: OpenSim 0.9.3.0 Nessie Dev 854f672 (X64/Unix/DotNet)physics ubODE 1.0-0.13.4
[12:30 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Speaking of changing things at runtime. I did get changing map tile renderer at runtime working quite well now. Still cleaning up the mess I made, but it happily switches between them now
[12:31 PM PT] Ubit Umarov: what map tile rendering? grid side ?
[12:31 PM PT] Vincent.Sylvester @hg.zetaworlds.com: What regions generate, warp3d or maptilerenderer
[12:32 PM PT] Ubit Umarov: oh and why do you want that? hihihi :)
[12:32 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Just silly experiments really
[12:32 PM PT] Ubit Umarov: a few weeks ago i tried to speedup warp3d a bit
[12:32 PM PT] Ubit Umarov: not much success .. :)
[12:32 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Decoupled it from startup so it doesn't block region boot up. Made it run in own thread so it stops blocking console. Silly really, but neat to have
[12:33 PM PT] Ubit Umarov: wel it is fast.. we just don't notice :)
[12:33 PM PT] Ubit Umarov: faster..
[12:33 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Retiring the V1 tiles to be part of scene data instead of assets, reduce that spam vector
[12:33 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Stuff doesn't work over hypergrid anyways so why bother
[12:34 PM PT] Ubit Umarov: well we do it just before region register on grid, so the map is there when registered
[12:34 PM PT] Ubit Umarov: in fact all that register into grid is plain BAD
[12:34 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Yeah but generating a new maptile effectively does a registration anyways so doesn't really matter if it is just water while it generates the map
[12:34 PM PT] Ubit Umarov: it should start a lot earlier before scene load, and possible witha login password
[12:35 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Decoupling also means technically you can generate a normal tile first and quickly and then go fancy warp3d
[12:35 PM PT] Ubit Umarov: then sending region state updates, including "im alive" ones
[12:35 PM PT] Ubit Umarov: in fact possible replacing the ugly datasnapshot module
[12:36 PM PT] Ubit Umarov: ( send less information that that mess )
[12:36 PM PT] Ubit Umarov: than..
[12:36 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Yeah don't remind me, that module is such a mess. I also think it's leaking memory or cpu somewhere somehow
[12:36 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I noticed on some regions after a couple weeks of runtime each snapshot run creates massive cpu spike
[12:37 PM PT] Gavin.Hird @grid.xmir.org:8002: speaking of which. what is the status on opensimsearch now?
[12:37 PM PT] Ubit Umarov: sadle the way comunications code is attached to main code, there is none avaiable at scene start
[12:37 PM PT] Ubit Umarov: a big nonsense..
[12:37 PM PT] Vincent.Sylvester @hg.zetaworlds.com: With Andrew away for now you can use the fork I made of it, which should work
[12:37 PM PT] Ubit Umarov: we have none on core, osgrid uses the one from Andrew
[12:38 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Granted it still has some of the bugs I found in rewriting it
[12:38 PM PT] Gavin.Hird @grid.xmir.org:8002: is there a location for the fork?
[12:38 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It should be linked on github, pull request the repo has is mine so you can quickly find it through that
[12:39 PM PT] Gavin.Hird @grid.xmir.org:8002: ok
[12:39 PM PT] Vincent.Sylvester @hg.zetaworlds.com: There are two bugs in the parsing that I found, but I can't recall what those were, would have to go through it again. There were also missing categories
[12:39 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Some of the events listing stuff was just wrong entirely in how it returned paginated results
[12:40 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I rewrote it in CI3 so the code doesn't directly work in plain php, but could be made to most likely
[12:40 PM PT] Vincent.Sylvester @hg.zetaworlds.com: What struck me most is that there is no way to administrate the results in terms of hiding results or approving them, so users can just post what they want in some of it and all you can do is just delete it from the database
[12:41 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Equally the structure we have for the data is a bit messy with the various tables, but that's also part of datasnapshot so changes need module changes too
[12:41 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Eventually I might just rewrite that part also and make it a bit more logical
[12:42 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Tables also have mismatching charset and collation which makes joining them for search results painfully slow unless you unify them
[12:43 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Probably the silliest thing though. Viewers, at least Firestorm, display the event times as local timezone, ignoring grid time entirely. In db they are stored as unix times so you have three timezones to fiddle with and it seems like the wrong approach to show events in local time and then mark them as upcoming at 2pm when grid time is already 6pm
[12:45 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Unclear where search is headed in the viewer. FS refuses to make changes to what they call legacy, but full web search is not really present either and just displaying a webpage is a bit silly in my opinion and it's quite a bit of work to get that working with the space the viewer allocates for that tab in search window
[12:46 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Nevermind running a full search engine with the functions people expect on this data is not easy
[12:47 PM PT] Vincent.Sylvester @hg.zetaworlds.com: If anyone is interested in more let me know, I'll shut up about search now :)
[12:47 PM PT] Gavin.Hird @grid.xmir.org:8002: I looked into it and most of the searches can be made by simple queries directly to the grid database
[12:48 PM PT] Gavin.Hird @grid.xmir.org:8002: some addition must be made, but not that many
[12:49 PM PT] Kayaker Magic: Speaking of searching the database directly, is there a better way to fetch assets like textures? Is there a region or robust request that checks the region cache first?
[12:50 PM PT] Gavin.Hird @grid.xmir.org:8002: who will fetch the asset?
[12:50 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Default cap should already go through cache first before region putting in a request with assets
[12:51 PM PT] Kayaker Magic: I'm fetching textures with a PHP script to find corrupt images, then trying to find what objects contain them.
[12:51 PM PT] Gavin.Hird @grid.xmir.org:8002: ok
[12:51 PM PT] Gavin.Hird @grid.xmir.org:8002: do you inspect the image for corruption or does the server spit out a message the texture is corrupt?
[12:52 PM PT] Kayaker Magic: I'm decoding the image and catching errors on that.
[12:52 PM PT] Gavin.Hird @grid.xmir.org:8002: so you go through all the images manually?
[12:52 PM PT] Vincent.Sylvester @hg.zetaworlds.com: They should not get corrupted in the first place though. Least I have not seen that happen for a long time
[12:53 PM PT] Kayaker Magic: I get errors on region start all the time!
[12:53 PM PT] Gavin.Hird @grid.xmir.org:8002: with a uuid for the corrupt image?
[12:53 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Some of that is dynamic texture references
[12:54 PM PT] Kayaker Magic: sometimes a UUID for the object, somettimes a uuid for the image, depending on the error.
[12:54 PM PT] Gavin.Hird @grid.xmir.org:8002: ok, yes I sometimes see the same
[12:54 PM PT] Cuga.Rajal @rajal.org:9000: I have textures in my region that throw console errors when an avatar's viewer gets close enoiugh to the object and requests
[12:54 PM PT] Kayaker Magic: I'd like to do a pass on the database and clean them all up.
[12:55 PM PT] Gavin.Hird @grid.xmir.org:8002: I use a tool called Navicat (paid for) to access teh db, and you can see the images for each record in the asset table
[12:55 PM PT] Kayaker Magic: I've got PHP web pages that do that.
[12:55 PM PT] Gavin.Hird @grid.xmir.org:8002: ok
[12:55 PM PT] Ubit Umarov: best way to get textures is to the use standard region service
[12:56 PM PT] Kayaker Magic: Using the Image Magick class for PHP, I think it also uses openjpeg
[12:56 PM PT] Ubit Umarov: you cluld bypass it and get from assets service directly, but would lose all dynamic textures
[12:56 PM PT] Kayaker Magic: Where is that standard region service documented?
[12:56 PM PT] Ubit Umarov: and avatars baked textures, possible
[12:57 PM PT] Ubit Umarov: ofc course it is documented.. on code :P
[12:57 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I don't recall textures going bad being a thing that happens often enough to require checking all databases. You sure those errors are showing actual corrupted objects?
[12:57 PM PT] Ubit Umarov: the region one also hads a little of protection
[12:57 PM PT] Ubit Umarov: because each viewer gets its own access URL
[12:57 PM PT] Kayaker Magic: OK, what code should I start reading?
[12:58 PM PT] Gavin.Hird @grid.xmir.org:8002: I think they got corrupted a couple years back on my part. May have been early code or some screwup I made - dunno
[12:58 PM PT] Gavin.Hird @grid.xmir.org:8002: have not seen corruption lately
[12:58 PM PT] Ubit Umarov: while exposed grid service will give you any asset, as long you know the UUID, like happens at SL now
[12:59 PM PT] Ubit Umarov: so funny... old MOSES once claimed we had a big security issuem because osgrid had exposed assets service and now so does SL.. fun fun
[12:59 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Some old assets might not work if they came from aurora or whitecore, but that would show up visibly as well, not just as console error
[12:59 PM PT] Cuga.Rajal @rajal.org:9000: I think my console errors for textures were result of not being found, not a corrupt image file
[1:00 PM PT] Ubit Umarov: we do give viewers several urls.. but service code is the same
[1:00 PM PT] Gavin.Hird @grid.xmir.org:8002: imports from old oars can also have corrupt textures
[1:01 PM PT] Ubit Umarov: ...OpenSim\Capabilities\Handlers\GetAssets\GetAssetsHandler.cs
[1:01 PM PT] Orbert.Tatham @hg.zetaworlds.com: Those imports aren't sanity checked before going into the database?
[1:01 PM PT] Vincent.Sylvester @hg.zetaworlds.com: hahaha
[1:01 PM PT] Vincent.Sylvester @hg.zetaworlds.com: nah
[1:01 PM PT] Ubit Umarov: this services all request for assets..
[1:02 PM PT] Ubit Umarov: ( well most of them )
[1:02 PM PT] Ubit Umarov: this also lets each region cache assets, taking load out of main grid services
[1:02 PM PT] Orbert.Tatham @hg.zetaworlds.com: Sorry, Vincent - I guess I am just naturally a much more paranoid coder than that.
[1:02 PM PT] Gavin.Hird @grid.xmir.org:8002: here is such a corruption message that is legit
[1:02 PM PT] Gavin.Hird @grid.xmir.org:8002: 75.333435, 26.561218>, exception EOF reached befor2023-04-10 20:51:57,984 DEBUG [WORLD2023-04-10 20:52:19,365 DEBUG [WORLD MAP]: Storing map image ba236f43-8d09-4b93-9a63-15cdb3362bad for Murat
[1:03 PM PT] Joe Magarac: Have to go, next meeting. Bye!
[1:03 PM PT] Ubit Umarov: cya joe
[1:03 PM PT] Vincent.Sylvester @hg.zetaworlds.com: That usually means it just got too large to process
[1:03 PM PT] Ubit Umarov: that is a bad jp2000
[1:04 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Don't know why and how that keeps happening, but on maptile create you can ignore it
[1:04 PM PT] Ubit Umarov: sounds those are 2 dif messages
[1:04 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It's not ideal, but figuring out why that might be is digging a hole through to china and back
[1:04 PM PT] Kayaker Magic: That's what everybody says about OpenSim errors: Just ignore it.
[1:04 PM PT] Ubit Umarov: first is about a bas prim face texture
[1:04 PM PT] Ubit Umarov: a bad..
[1:05 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Well in case of textures if you actually see a prim going white or gray from texture corruption, otherwise most of the missing or corrupt texture debug log is just random bake data or dynamic textures
[1:05 PM PT] Vincent.Sylvester @hg.zetaworlds.com: For textures to go bad something would have to interact with the data itself, so assets being updated
[1:05 PM PT] Gavin.Hird @grid.xmir.org:8002: another
[1:05 PM PT] Gavin.Hird @grid.xmir.org:8002: 2023-04-17 14:05:05,687 WARN [Warp3D]: Failed to decode texture f828406b-7ec9-48fd-9650-8b849e332d30 for prim meshFoodBasket at , exception EOF reached before finding Contiguous Codestream Box
2023-04-19 10:19:46,845 WARN [Warp3D]: Failed to decode texture f828406b-7ec9-48fd-965
[1:06 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Warp3d does that a lot
[1:06 PM PT] Ubit Umarov: wel message tells it all :)
[1:06 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Even an empty region it'll say that if you make it large enough lol
[1:06 PM PT] Ubit Umarov: the decoding of the jpeg is failing
[1:06 PM PT] Gavin.Hird @grid.xmir.org:8002: yes
[1:07 PM PT] Ubit Umarov: you can try to tell it to use openjpeg
[1:07 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It pulls data together and tries to create a picture out of that and then falls over and throws a tantrum when it gets too complicated
[1:07 PM PT] Cuga.Rajal @rajal.org:9000: You know the prim name meshFoodBasket so you can look up the object in the db
[1:07 PM PT] Gavin.Hird @grid.xmir.org:8002: 2023-02-08 10:35:24,768 WARN [J2KDecoderModule]: CSJ2K threw an exception decoding texture 6fcfc424-1a22-41ae-b139-1b9c3c7d5774: EOF reached before finding Contiguous Codestream Box
[1:07 PM PT] Ubit Umarov: but default is uses CSJ2K
[1:07 PM PT] Gavin.Hird @grid.xmir.org:8002: yes Cuga
[1:07 PM PT] Ubit Umarov: by default..
[1:08 PM PT] Ubit Umarov: but that is it,, a bad texture or some that CSJ2k could not decode
[1:09 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Warp3d attempts to apply textures to things, but when the visible image becomes complex or it has to bend textures, stuff like that, it struggles. You can break it just by changing the terrain texture heights even
[1:09 PM PT] Ubit Umarov: on opensim.ini
[1:09 PM PT] Ubit Umarov: ; Image decoding. Use CSJ2K for layer boundary decoding if true,
; OpenJPEG if false
; UseCSJ2K = true
[1:09 PM PT] Ubit Umarov: if you set to false it will use openjpeg
[1:10 PM PT] Ubit Umarov: hmm i think :)
[1:10 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Now whether that works better remains to be seen
[1:11 PM PT] Ubit Umarov: ok any more issue for today?
[1:11 PM PT] Vincent.Sylvester @hg.zetaworlds.com: You are throwing it a curve ball really. It attempts to generate or render a picture without actually simulating things, just via the data it has available to it. It's like rendering a scene by just being told "There is a lovely tree on the right with green leaves and some fruits on it" heh
[1:11 PM PT] Gavin.Hird @grid.xmir.org:8002: no, I need to go any minute
[1:11 PM PT] Ubit Umarov: in fact if does not render ll trees :p
[1:12 PM PT] Ubit Umarov: but well those have not fruits :p
[1:12 PM PT] Vincent.Sylvester @hg.zetaworlds.com: lol
[1:12 PM PT] Gavin.Hird @grid.xmir.org:8002: which is a good thing as they are butt ugly
[1:12 PM PT] Ubit Umarov: sadly
[1:12 PM PT] Ubit Umarov: they are important things
[1:12 PM PT] Lyr Lobo grins
[1:12 PM PT] Ubit Umarov: them and grass
[1:12 PM PT] Motoko.Karu @login.digiworldz.com:8002: Gotta run.. have a great week everyone
[1:12 PM PT] Cuga.Rajal @rajal.org:9000: somebbody here talked about freshening them up :)
[1:13 PM PT] Ubit Umarov: but ofc 120% viewer side
[1:13 PM PT] Gavin.Hird @grid.xmir.org:8002: actually some LL regions have a newer version of LL trees
[1:13 PM PT] Ubit Umarov: well it is viewer side
[1:13 PM PT] Kayaker Magic: I think warp3D does a great job, I just only let it run late at nite or by request.
[1:14 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Tree It could be used for making new ones, it has a fairly permissive license and the trees look quite nice
[1:14 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Thing is that's mostly viewer side anyways
[1:14 PM PT] Orbert.Tatham @hg.zetaworlds.com: "Someone here" gave up becuase nobody was interested
[1:14 PM PT] Cuga.Rajal @rajal.org:9000: yep :)
[1:14 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Be neat if we could send them over as well, but that be a bit much
[1:15 PM PT] Ubit Umarov: region only guess some parameters that define the vegetable
[1:15 PM PT] Cuga.Rajal @rajal.org:9000: yeah you have to herd cats to take that one on
[1:15 PM PT] Ubit Umarov: outch... my typing is so good.. i meant region only gets and sends...
[1:15 PM PT] Cuga.Rajal @rajal.org:9000: client and voewer
[1:15 PM PT] Gavin.Hird @grid.xmir.org:8002: I need to go
[1:15 PM PT] Gavin.Hird @grid.xmir.org:8002: cheers
[1:15 PM PT] Lyr Lobo: bye gavin
[1:15 PM PT] Lyr Lobo: great to see you
[1:16 PM PT] Ubit Umarov: wel i also need to go
[1:16 PM PT] Cuga.Rajal @rajal.org:9000: tc Gavin
[1:16 PM PT] Lyr Lobo: Have a great day
[1:16 PM PT] Cuga.Rajal @rajal.org:9000: I need to as well
[1:16 PM PT] Lyr Lobo: Hi Cuga *grins*
[1:16 PM PT] Kayaker Magic: So is this the end of meeting?
[1:16 PM PT] Ubit Umarov: thanks and hope to see you all next week if nt before :)
[1:16 PM PT] Lyr Lobo: Yes, me too..take care, everyone
[1:16 PM PT] Cuga.Rajal @rajal.org:9000: grins back
[1:16 PM PT] Lyr Lobo: Heya Kayaker!
[1:16 PM PT] Cuga.Rajal @rajal.org:9000: take care all
