Chat log from the meeting on 2024-03-05

From OpenSimulator

Jump to: navigation, search
[12:14 PM PT]  Ubit Umarov: welcome :)
[12:14 PM PT]  Kayaker Magic: I'll repeat:  I have questions about how owner and creator get assigned on OAR loads...
[12:14 PM PT]  Ubit Umarov: just short code changes
[12:14 PM PT]  Kayaker Magic: I loaded an OAR on a new grid, all the items in it got "owned" by the owner of the region, the creators got correctly assigned to people from other grids, even people who had never visited the new grid. How did the HG creator names get found?
[12:14 PM PT]  Vincent.Sylvester The creator data is saved as UUI which is the uuid and the grid url combined
[12:14 PM PT]  Vincent.Sylvester With that you have all you need to query them for profile data including pictures
[12:15 PM PT]  Ubit Umarov: i did change the nre functions  osget*InventoryKey to osget*inventoryItemKey
[12:15 PM PT]  Kayaker Magic: Then the OAR load calls up the remote grid and gets the avatar name?
[12:15 PM PT]  Ubit Umarov: do match the older ones
[12:15 PM PT]  Vincent.Sylvester Yep
[12:15 PM PT]  Kayaker Magic: Assuming the other grid is up and responding....
[12:15 PM PT]  Ubit Umarov: and to make clear that the key returned is the item key, not its asset key like on the ll one
[12:16 PM PT]  Vincent.Sylvester The name is already included in uui, it's uuid, name and grid url
[12:16 PM PT]  Vincent.Sylvester Only picture is fetched along with some profile data
[12:16 PM PT]  Kayaker Magic: But if the OAR was made on that grid, does the OAR save still save uuid, name and grid url?
[12:17 PM PT]  Ubit Umarov: UUI and includes the name
[12:17 PM PT]  Ubit Umarov: also..
[12:17 PM PT]  Ubit Umarov: creator data should always be saved..
[12:17 PM PT]  Kayaker Magic: I hope so!
[12:17 PM PT]  Ubit Umarov: sadly seems to not happen on same cases
[12:17 PM PT]  Vincent.Sylvester Interesting question. It very well might save all as uui, though I am not sure, would have to look
[12:17 PM PT]  Vincent.Sylvester Possible locally it doesn't
[12:18 PM PT]  Kayaker Magic: So to be clear: OAR save always saves name and URL with creator UUID? Does it do that with owner UUIDs?
[12:18 PM PT]  Ubit Umarov: owner may be replaced
[12:19 PM PT]  Ubit Umarov: guess it is by default??
[12:19 PM PT]  Vincent.Sylvester I don't know if it always saves UUI for creator, certainly if the items are already remote made in the first place
[12:19 PM PT]  Vincent.Sylvester OARs are designed for backups less for sharing
[12:19 PM PT]  Ubit Umarov: creator data is stored if it is avaiable
[12:19 PM PT]  Kayaker Magic: This OAR came from Kitely, they may have modified OAR save to save URI for creator if normal OpenSim does not....
[12:20 PM PT]  Ubit Umarov: well problem witn oars wa that they where made for a pseudo free world
[12:20 PM PT]  Ubit Umarov: so all data can be lost or changed
[12:20 PM PT]  Kayaker Magic: Understood
[12:21 PM PT]  Ubit Umarov: something impossible to fix on a opensource code using xml crap
[12:21 PM PT]  Kayaker Magic: But I want to know best practice for preserving creator whenever possible.
[12:21 PM PT]  Ubit Umarov: (well even woth close source code.. )
[12:22 PM PT]  Ubit Umarov: current code code does try to preserve creator, ofc
[12:22 PM PT]  Ubit Umarov: core code..
[12:22 PM PT]  Vincent.Sylvester Likely that for local creator it only stores uuid and so when an oar is moved the data doesn't match a local account, still looking at the code though gimme a minute
[12:24 PM PT]  Ubit Umarov: oars  created on a non HG grid may not have all creator info
[12:24 PM PT]  Ubit Umarov: or standalone
[12:24 PM PT]  Kayaker Magic: I see the documentation (not the source!) does say that the estate owner becomes the owner of all objects if --default-user is not specified.
[12:25 PM PT]  Ubit Umarov: duhh found my fs issue on your textures
[12:26 PM PT]  Ubit Umarov: it was on minimal graphics settings
[12:26 PM PT]  Ubit Umarov: when i did update it, it crashed on first run and did that...
[12:26 PM PT]  Vincent.Sylvester sop has a field for the creatordata, but I think sog does not and derives that from rootpart
[12:27 PM PT]  Vincent.Sylvester So it might be possible that when moving grids via OAR some data is lost
[12:27 PM PT]  Ubit Umarov: no
[12:27 PM PT]  Ubit Umarov: think creator data is kept and stored per prim
[12:27 PM PT]  Ubit Umarov: and should be like that
[12:28 PM PT]  Ubit Umarov: in theory once could build objects made with parts created by others..
[12:28 PM PT]  Ubit Umarov: so that should be kept somewhere
[12:29 PM PT]  Vincent.Sylvester It should, but sog itself has no field for creatordata, but for creator uuid, which is a bit odd
[12:29 PM PT]  Ubit Umarov: on that sog gets it from root
[12:30 PM PT]  Ubit Umarov: Sl has no SOG,.. everything is on root part
[12:30 PM PT]  Ubit Umarov: im changing that..
[12:30 PM PT]  Ubit Umarov: for now we do the same on most things..
[12:31 PM PT]  Ubit Umarov: that means that all prims have fileds for things that are only valid for SOG
[12:31 PM PT]  Ubit Umarov: so ill move some of those to SOG; since we do have it
[12:31 PM PT]  Joe Magarac: SOG?
[12:31 PM PT]  Vincent.Sylvester scene object group
[12:31 PM PT]  Ubit Umarov: btu creator data is not that case
[12:31 PM PT]  Ubit Umarov: SOG == linkset
[12:32 PM PT]  Joe Magarac: Ah.
[12:32 PM PT]  Ubit Umarov: keyframe data for example is stored on SOG part of oars.. not parts
[12:33 PM PT]  Ubit Umarov: sadly dbs do not know abotu sog, so it is stored on root still
[12:33 PM PT]  Vincent.Sylvester Only thing that modifies creator data is home parameter for save oar command, but it should store all so maybe the restore function doesn't properly deserialize it dunno
[12:33 PM PT]  Ubit Umarov: things TODO :)
[12:35 PM PT]  Ubit Umarov: remember.. some coredevs had other ideas about IP rights
[12:35 PM PT]  Vincent.Sylvester Only way save would fail is if the creator data is empty and no home parameter is passed to save oar, then that falls through
[12:35 PM PT]  Ubit Umarov: in fact.. some may had been totally against any IP rights
[12:35 PM PT]  Ubit Umarov: so their code is possible arounf
[12:35 PM PT]  Ubit Umarov: around
[12:36 PM PT]  Ubit Umarov: like options to override all original creator data..
[12:37 PM PT]  Orbert.Tatham It is easy to do at the server level anyway
[12:37 PM PT]  Vincent.Sylvester Be neat if that really is a bug if we could get some oars that are "broken" to reproduce and check what the serializer sees
[12:37 PM PT]  Ubit Umarov: (ofc regions may run own "forks" with own ideas abotu that )
[12:37 PM PT]  Kayaker Magic: Not VERY easy Orbert, linkset creator data is hidden in XML in linkset blob in the database...
[12:38 PM PT]  Ubit Umarov: should be hidden even more ..  but impossible
[12:38 PM PT]  Ubit Umarov: reason why ppl talk abotu blockchains blabla..
[12:39 PM PT]  Ubit Umarov: equaly useless..  just more "fancy"
[12:39 PM PT]  Ubit Umarov: (in same cases just totaly impratical (
[12:39 PM PT]  Orbert.Tatham If you seriously want to protect the legal fiction that there is such a thing as intellectual property, go back to SL
[12:39 PM PT]  Orbert.Tatham You can't get into there servers
[12:40 PM PT]  Orbert.Tatham *their
[12:40 PM PT]  Joe Magarac: I used to go to that metaverse standards meeting. Nothing came of that. There was a good idea for identifying near-duplicate meshes, though.
[12:40 PM PT]  Ubit Umarov: well i remember the case of CRT tube and RCA  full violation of the patents
[12:40 PM PT]  Ubit Umarov: lost in court.. even so, never paid it
[12:41 PM PT]  Orbert.Tatham You can't rewrite the source code of the patent, either
[12:41 PM PT]  Ubit Umarov: those near dup meshs are not easy either
[12:42 PM PT]  Ubit Umarov: even if you do get the right parts of the asset to compare..
[12:42 PM PT]  Joe Magarac: No, RCA hired Zworklin and he did OK.
[12:42 PM PT]  Orbert.Tatham From an IP standpoint, a near-duplicate is still unique
[12:42 PM PT]  Ubit Umarov: crt patent did belong to other
[12:43 PM PT]  Joe Magarac: I could argue about Farnsworth, but not here.
[12:43 PM PT]  Ubit Umarov: >(
[12:43 PM PT]  Joe Magarac: Anyway, before we run out of time, there's a new version of Sharpview out:
[12:43 PM PT]  Ubit Umarov: Farnsworth did own in court.. even so rca never paid
[12:43 PM PT]  Ubit Umarov: and a lot more examples
[12:44 PM PT]  Ubit Umarov: ty
[12:44 PM PT]  Ubit Umarov: did won i meant
[12:44 PM PT]  Joe Magarac: To download, username "devs", password "thread".
[12:44 PM PT]  Ubit Umarov: ty :)
[12:44 PM PT]  Joe Magarac: Use with a low-value alt.
[12:45 PM PT]  Kayaker Magic: You have a Linux version of that Joe?
[12:45 PM PT]  Joe Magarac: Yes.
[12:45 PM PT]  Ubit Umarov: on code changes i did fix a silly bug on get*params
[12:45 PM PT]  Joe Magarac: Linux and Windows.
[12:45 PM PT]  Ubit Umarov: mating 9118
[12:45 PM PT]  Ubit Umarov: yeack.. mantis 9118!
[12:46 PM PT]  Lyr Lobo: Hello *smiles*  back from the dentist
[12:46 PM PT]  Ubit Umarov: welcome..
[12:46 PM PT]  Lyr Lobo: thanks
[12:46 PM PT]  Ubit Umarov: good to see you stil have teeth
[12:46 PM PT]  Ubit Umarov: ;(
[12:46 PM PT]  Ubit Umarov: oops :)
[12:46 PM PT]  Lyr Lobo: nice new crows and all comfy *grins*
[12:46 PM PT]  Kayaker Magic: My dentist office is called "Smile Country"
[12:47 PM PT]  Ubit Umarov: we also did add osGetSitTarget
[12:47 PM PT]  Lyr Lobo chuckles and nods
[12:47 PM PT]  Vincent.Sylvester The xml to sop code is just "read line by line" basically so that should definitely restore the data as it was written, weird
[12:47 PM PT]  Lyr Lobo: mine is high tech and stylish with comfort key
[12:47 PM PT]  Orbert.Tatham I only smile in the dentist's office with enough laughing gas in me
[12:47 PM PT]  Joe Magarac: This version of Sharpview can do region crossings. But it needs a recent Open Simulator. Ubittest  Ubittest2 works perfectly. Other places may need an upgrade.
[12:47 PM PT]  Vincent.Sylvester No I think this should work fine unless creator data itself is not written, which it should be
[12:48 PM PT]  Joe Magarac: I'd like to ask that the simulator version number be increased. There are too many different versions of
[12:49 PM PT]  Vincent.Sylvester That's not reliable metric I'm afraid
[12:49 PM PT]  Vincent.Sylvester Cause anyone can change that fairly easily
[12:49 PM PT]  Joe Magarac: That's true of any open source program.
[12:49 PM PT]  Joe Magarac: But few people change the numbers.
[12:50 PM PT]  Joe Magarac: Sharpview will fail to do region crossings on some simulators that say but are different builds.
[12:51 PM PT]  Vincent.Sylvester You know what they say about assumptions
[12:51 PM PT]  Kayaker Magic: I assume we will asymptotically approach version
[12:51 PM PT]  Vincent.Sylvester Only real reliable metric I have found is to "poke" for certain functions and things that are only present in specific versions
[12:51 PM PT]  Vincent.Sylvester Parsing version strings a can of worms
[12:51 PM PT]  Joe Magarac: But changing them is not.
[12:52 PM PT]  Joe Magarac: Is this a religious issue? Usually this isn't a big problem.
[12:52 PM PT]  Ubit Umarov: joe it is Dev.. needs the commit number for full spec
[12:52 PM PT]  Vincent.Sylvester I just know that it's one of the first things folks usually do when they fork
[12:52 PM PT]  Cuga.Rajal neighboring regions don't send "capabilities" like is sent on TP?
[12:53 PM PT]  Joe Magarac: But I can't tell if something is later or earlier from the commit ID.
[12:53 PM PT]  Joe Magarac: Who decdes what OSGrid runs as standard?
[12:53 PM PT]  Cuga.Rajal you can look up the commit date, but that is a PITA
[12:54 PM PT]  Joe Magarac: Look it up where? Can't be sure what fork it comes from.
[12:54 PM PT]  Cuga.Rajal I think theres a git command that will tell you
[12:55 PM PT]  Cuga.Rajal still, not something that can be programmed into a biewer
[12:55 PM PT]  Cuga.Rajal or automated
[12:55 PM PT]  Joe Magarac: Here's my position. I am willing to work to make Sharpview compatible with Open Simulator. But I need to know what version of Open Simulator I need to be compatible with.
[12:55 PM PT]  Orbert.Tatham Moving target problem
[12:56 PM PT]  Joe Magarac: If you guys can't get your version act together, I will not be able to do much about compatibility,.
[12:56 PM PT]  Cuga.Rajal I think Joe's comment is valid, tho I donlt know the solution
[12:56 PM PT]  Vincent.Sylvester With the master version ideally, but at least the last release. Thing is you cannot tell by version string what you are dealing with. That said you also don't have to care either. If it works with release and master dev and older versions just simply don't work that's on the region operator to fix
[12:57 PM PT]  Vincent.Sylvester People keeping their stuff on wholly outdated versions only have themselves to blame
[12:58 PM PT]  Joe Magarac: Currently, I only support OSGrid-affiliated simulators, because that provides some standarization.
[12:58 PM PT]  Kayaker Magic: OSGrid? I hear rumors that there are regions out there still on 0.6 of OpenSim
[12:58 PM PT]  Ubit Umarov: osgrid adms decided that..  currently that follow our code
[12:58 PM PT]  Joe Magarac: Anyway, I don't need an answer right now. But think about this. I'll have a talk with Beq Janus on how she decides what to support.
[12:59 PM PT]  Ubit Umarov: btu osg onlu decides their own regions and code distro
[12:59 PM PT]  Ubit Umarov: so yeah possilble some 0.7 still try to connect
[1:00 PM PT]  Joe Magarac: Anyway, I'll leave you guys to sort that one out. Off to next meeting. See you next week! Have fun with Sharpview. If it breaks, mail "".
[1:00 PM PT]  Kayaker Magic: Yeah, it is noon here, I have chores in RL to get to.
[1:00 PM PT]  Kayaker Magic: So buy all!
[1:00 PM PT]  Orbert.Tatham Peace, Kayaker'
[1:01 PM PT]  Cuga.Rajal I came in late, was there a Sharpview release?
[1:01 PM PT]  animatsalt resident: Bye.
[1:01 PM PT]  Vincent.Sylvester In other news. After poking git for three days rebasing has worked so the road to new xunit tests and checking migrations is clear
[1:01 PM PT]  Vincent.Sylvester I have to check the tests again and see if they run fine on nix and win, but they did last time
[1:02 PM PT]  Vincent.Sylvester Unfortunately prebuild is useless so will have to ship csproj file along with them
[1:02 PM PT]  Vincent.Sylvester Fortunately they are barren files that don't contain any platform specific things far as I can tell
[1:03 PM PT]  Ubit Umarov: prebuild is not useless :p
[1:03 PM PT]  Ubit Umarov: direct use of project files can be a disaster
[1:03 PM PT]  Vincent.Sylvester Well useless for writing csproj files that are test projects, it doesn't know how to write those
[1:04 PM PT]  Vincent.Sylvester If I add prebuild it just writes a project file instead of a test project file, so the tests fail
[1:04 PM PT]  Ubit Umarov: you do need to add the test project also
[1:04 PM PT]  Ubit Umarov: to the xml
[1:04 PM PT]  Ubit Umarov: think you mean unit test
[1:05 PM PT]  Vincent.Sylvester I can't, because prebuild then writes it wrong, it doesn't know how to write that specific type of project file
[1:05 PM PT]  Vincent.Sylvester It just writes a normal csproj as all the rest, but it needs to write a test project not a project
[1:05 PM PT]  Ubit Umarov: wel it did for all the unit 2.0 tests
[1:05 PM PT]  Ubit Umarov: so it may need fix for more recent ones
[1:05 PM PT]  Vincent.Sylvester Cause those were different structure and just based on a self-contained app, xunit runs directly in dotnet
[1:06 PM PT]  Ubit Umarov: it was set for unit 2.0
[1:06 PM PT]  Ubit Umarov: ofc does not cover x...
[1:06 PM PT]  Ubit Umarov: we don't eitehr
[1:07 PM PT]  Ubit Umarov: well still
[1:07 PM PT]  Cuga.Rajal brb
[1:07 PM PT]  Vincent.Sylvester The project files it wrote for nunit were same as normal projects, it didn't write them any different
[1:07 PM PT]  Vincent.Sylvester xunit requires test projects
[1:07 PM PT]  Ubit Umarov: several projects did remove prebuild
[1:07 PM PT]  Vincent.Sylvester Which have different structure
[1:07 PM PT]  Ubit Umarov: but as i said, it is still usefull.. just needs more work
[1:08 PM PT]  Vincent.Sylvester The project file xunit needs though is just some references to local files which are platform independent
[1:08 PM PT]  Ubit Umarov: for example on libomv is is bad
[1:08 PM PT]  Ubit Umarov: bc ms stupidity
[1:08 PM PT]  Ubit Umarov: as you know there is nto just dotnetX...
[1:08 PM PT]  Vincent.Sylvester So in this case shipping csproj might not cause any issues, at least I don't see how it would with no platform specific references in there
[1:09 PM PT]  Ubit Umarov: there are dotnetx -- dotnetx-linux etvc
[1:09 PM PT]  Ubit Umarov: and several libomv projects only work on windows now
[1:09 PM PT]  Ubit Umarov: need to be set for dotnet6-windows
[1:09 PM PT]  Ubit Umarov: and also need stupid flags to allow several components
[1:10 PM PT]  Vincent.Sylvester I got jenkins to build it and the test project so it compiles platform independent which leads me to think it'll probably work to run the tests on linux as well, they certainly work fine on windows
[1:10 PM PT]  Vincent.Sylvester I can't test mac
[1:10 PM PT]  Ubit Umarov: ewll diferent problems
[1:10 PM PT]  Ubit Umarov: that are also a huge pain messing with project files directly
[1:11 PM PT]  Cuga.Rajal I can help w/Mac if I know what to do
[1:11 PM PT]  Vincent.Sylvester Which is why I'm saying it's a blessing test projects are fairly simple so it shouldn't cause platform problems
[1:12 PM PT]  Vincent.Sylvester I'm going to refine that a bit more in the coming weeks and write a few tests that actually attempt do more than just check if OpenSim.dll exists
[1:12 PM PT]  Vincent.Sylvester That should show whether this works, but I haven't seen any major red flag
[1:12 PM PT]  Vincent.Sylvester I also finished a migrations check routine that checks the existing database tables against a known good schema and throws a tantrum if they diverge
[1:13 PM PT]  Vincent.Sylvester This is to prevent any migration snafu from breaking stuff and alerting the user that maybe they need to manually fix their tables
[1:13 PM PT]  Vincent.Sylvester Of course if they run custom stuff the schema files are just in bin so can be edited directly
[1:14 PM PT]  Vincent.Sylvester Currently the routine runs at the end of each migration to serve as final check if everything is as it should be before continuing startup
[1:15 PM PT]  Vincent.Sylvester It was a bit weird that we currently solely rely on some sql to work without any way to check if it actually did and the migrations just whizz by during startup without looking at logs you got no clue if they worked
[1:19 PM PT]  Cuga.Rajal I need to run.. take care everybody
[1:19 PM PT]  Orbert.Tatham Laterz, Cuga
[1:19 PM PT]  Ubit Umarov: ok guess time to close meeting log :)
[1:19 PM PT]  Ubit Umarov: hope to see  you all next week :)
Personal tools
About This Wiki