Chat log from the meeting on 2023-06-20

From OpenSimulator

Revision as of 12:07, 20 June 2023 by Tampa (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
[11:01 AM PT]  Andrew Hellershanks: Hello, everyone.
[11:02 AM PT]  Orbert.Tatham Andrew
[11:03 AM PT]  Selby.Evans hi everyone
[11:03 AM PT]  Tess JL: Hello Selby
[11:04 AM PT]  Lyr Lobo: Hello *smiles*
[11:05 AM PT]  Tess JL: Hi Lyr :-)
[11:05 AM PT]  Vincent.Sylvester Got no code changes this week, only more bugs to be fixed in future
[11:06 AM PT]  Lyr Lobo: Hi Tess *grins*
[11:06 AM PT]  Ubit Umarov: hi
[11:06 AM PT]  Lyr Lobo: Hi Ubit
[11:07 AM PT]  Tess JL: I suppose even developers deserve a break in summer. ;-)
[11:07 AM PT]  Andrew Hellershanks: There was a change to llGetObjectPrimCount() that makes it return 0 on attachments.  That is to make it conform to the specs.
[11:07 AM PT]  Andrew Hellershanks: Two new functions were added: osGetPrimCount([key primKey]) and osGetSittingAvatarsCount([key primkey]).
[11:08 AM PT]  Vincent.Sylvester Oh that, yeah minor silly things
[11:08 AM PT]  Andrew Hellershanks: That's all the changes for this past week.
[11:08 AM PT]  Andrew Hellershanks: Minor changes but I don't know how silly they are.
[11:08 AM PT]  Andrew Hellershanks: :)
[11:08 AM PT]  Vincent.Sylvester Well LL is silly and we try to make sense of it somehow
[11:09 AM PT]  Andrew Hellershanks that makes me think of Monty Python
[11:09 AM PT]  Andrew Hellershanks: Yes, we can but try. :)
[11:09 AM PT]  Orbert.Tatham LL  Monty Python? Roughly one as serious as the other
[11:09 AM PT]  Tess JL: How does osGetSittingAvatarsCount([key primkey]) handle multiple avatars sitting on the same prim?
[11:09 AM PT]  Ubit Umarov: silly things?
[11:09 AM PT]  Ubit Umarov: :p
[11:10 AM PT]  Vincent.Sylvester Silly that they are kinda needed in the first place cause heaven forbid LSL make sense
[11:10 AM PT]  Ubit Umarov: avatars don't sit on prims tess
[11:10 AM PT]  Ubit Umarov: well not completly
[11:11 AM PT]  Ubit Umarov: anyways we do have a list of all sitting on a linkset that returns its count
[11:11 AM PT]  Andrew Hellershanks: IIRC, avatars sitting on a prim/linkset are sort of added to the end of the linkset.
[11:12 AM PT]  Ubit Umarov: never where on our code
[11:12 AM PT]  Ubit Umarov: they are added to own list on the scenegroup
[11:12 AM PT]  Tess JL: Ok. The reason I asked is that SL has can be a bit funny if/when there are more sitting avatars than there are parts of the linkset they are sitting on. I didn't kwno if that applied to opensim though
[11:12 AM PT]  Ubit Umarov: to fast to get those 2 counts
[11:13 AM PT]  Andrew Hellershanks: Either I had seen that in the code ages ago when I was looking at the SOG/SOP code or I read it somewhere.
[11:13 AM PT]  Ubit Umarov: for most things avatars do sit on root prim..
[11:13 AM PT]  Ubit Umarov: ofc in detaisl they do get the actuall prim rotation etc..
[11:14 AM PT]  Jamie.Jordan hi everybody
[11:14 AM PT]  Andrew Hellershanks: Hello, Jamie
[11:14 AM PT]  Vincent.Sylvester Was working on multi sit script and asked a harmless question and of course Ubit finds and fixes a bug hehe
[11:15 AM PT]  Ubit Umarov: as sl, they go get linknumbers after the last linked prim
[11:15 AM PT]  Ubit Umarov: well i did decide to add those 2 ossl ones  that are clear and direct
[11:16 AM PT]  Ubit Umarov: unlike the several ll ones that are not coerent
[11:16 AM PT]  Andrew Hellershanks: That's what I think I had seen in the SOG code. They get link numbers.
[11:16 AM PT]  Ubit Umarov: yes andrew as spec
[11:16 AM PT]  Vincent.Sylvester LL and coherent, that's a parallel universe sadly
[11:17 AM PT]  Joe Magarac: Link numbers as shown in the viewer may not agree with link numbers known to the server.
[11:17 AM PT]  Ubit Umarov: for example LlGetNumberOfPrims  counts sitting avatars as prims
[11:17 AM PT]  Vincent.Sylvester That's been a long standing thing, there is even a recent mantis about this
[11:17 AM PT]  Ubit Umarov: llGetObjectPrimCount does not, but does not work on attachments
[11:18 AM PT]  Ubit Umarov: ( and you need the linkset uuid )
[11:18 AM PT]  Tess JL: So with osGetPrimCount() you can use the UUID of any part of the linkset?
[11:19 AM PT]  Tess JL: And even call it from outside the linkset?
[11:19 AM PT]  Ubit Umarov: not that i abused and used C# method selection by arguments types
[11:19 AM PT]  Ubit Umarov: not first time..  may make viewers script syntax a bit confused
[11:20 AM PT]  Vincent.Sylvester Speaking of syntax, need to update the xml as well
[11:20 AM PT]  Ubit Umarov: ie i added integer osGetPrimCount()
integer osGetPrimCount(key objectID)
[11:21 AM PT]  Ubit Umarov: ll whould had added
[11:21 AM PT]  Ubit Umarov: integer osGetPrimCount()
[11:21 AM PT]  Lyr Lobo: Yes the avatars counting as the sit count was important under Havok 4 and we are compatible, right?
[11:21 AM PT]  Ubit Umarov: integer osGetLinkPrimCount(key objectID)
[11:21 AM PT]  Lyr Lobo: while not under Havok, we stay in line with SL features
[11:21 AM PT]  Lyr Lobo: is what I mean
[11:22 AM PT]  Ubit Umarov: hmm no relation with havoc Lyr
[11:22 AM PT]  Lyr Lobo: right
[11:22 AM PT]  Ubit Umarov: in lsl we can do things to sitting avatars as if sitting prims
[11:22 AM PT]  Ubit Umarov: i mean as if child prims :)
[11:23 AM PT]  Ubit Umarov: like change sitting offset, while sitting
[11:23 AM PT]  Ubit Umarov: etc
[11:23 AM PT]  Ubit Umarov: physics here do not see avatars
[11:23 AM PT]  Cuga.Rajal I have a question about sitting avatars at the appropriate point
[11:24 AM PT]  Ubit Umarov: at sl tehy may see on collisions with ground or rays
[11:24 AM PT]  Ubit Umarov: i mean do not see sitting avatars..
[11:24 AM PT]  Ubit Umarov:
[11:25 AM PT]  Ubit Umarov:
[11:25 AM PT]  Joe Magarac: I've proposed a major but backwards compatible change to the prim hierarchy on the SL side. It may be implemented over there.
[11:25 AM PT]  Cuga.Rajal when an avatar sits on a vehicle, is the avatar "physics mesh" treated like physics type none?
[11:25 AM PT]  Joe Magarac: Comments from OS would be helpful.
[11:26 AM PT]  Joe Magarac: In SL, the avatar's physics mesh becomes part of the ilnkset.
[11:26 AM PT]  Ubit Umarov: they did not implemented it ?
[11:26 AM PT]  Joe Magarac: This does not include attachments.
[11:26 AM PT]  Vincent.Sylvester Would help if that page was actually public and not locked behind a login Joe
[11:26 AM PT]  Vincent.Sylvester You can propose changes on the mantis or mailing list too
[11:26 AM PT]  Ubit Umarov: well we have nothing abotu pivot points
[11:27 AM PT]  Joe Magarac: Oh, the SL JIRA is not even public to read. I'll copy that elsewhere.
[11:27 AM PT]  Ubit Umarov: .. UDP messages which send prim positions
to the viewer work now...
[11:27 AM PT]  Ubit Umarov: no they do not
[11:28 AM PT]  Ubit Umarov: udp only send a position...
[11:28 AM PT]  Cuga.Rajal hi, sorry late
[11:28 AM PT]  Ubit Umarov: you need to know if it is a root prim or a child one to transform
[11:28 AM PT]  Tess JL: It is marked as Public, Jow
[11:29 AM PT]  Ubit Umarov: and No... objects are not sent on ObjectUpdateCompressedObjectData
[11:29 AM PT]  Ubit Umarov: only objects that are cachable
[11:29 AM PT]  Ubit Umarov: others are sent as full object update
[11:30 AM PT]  Ubit Umarov: ( similar data... in fact better on older format..  but details )
[11:31 AM PT]  Ubit Umarov: and avatars do use variants of those
[11:31 AM PT]  Ubit Umarov: right?
[11:31 AM PT]  Ubit Umarov: ( this is in reference to the ll jira above )
[11:33 AM PT]  Ubit Umarov: yeah joe object hiearchy is more complex that the protocols
[11:33 AM PT]  Ubit Umarov: everything assumes the falt model   avatars
[11:34 AM PT]  Ubit Umarov: flat
[11:34 AM PT]  Ubit Umarov: personally i whould not add it as a hierachy of prims
[11:35 AM PT]  Ubit Umarov: but as a hierachy of linksets
[11:35 AM PT]  Joe Magarac: That would be a bigger change.
[11:35 AM PT]  Ubit Umarov: adding a higher level struct lets call i model  for example
[11:35 AM PT]  Joe Magarac: The existing protocol can represent a hierarchy.
[11:36 AM PT]  Tess JL: What would be the advantages of that solution, Ubit?
[11:36 AM PT]  Joe Magarac: It does for avatars on vehicles. Vehicle->Avatar->Attachment.
[11:36 AM PT]  Vincent.Sylvester That idea has been floating around for ages:
[11:36 AM PT]  Ubit Umarov: the model could then add linksets whit hiierachy connected at physics level by joints etc
[11:37 AM PT]  Ubit Umarov: so older ones cpould still be used as part of the modle
[11:37 AM PT]  Ubit Umarov: and a linkset whoul keep current static relative positions etc
[11:38 AM PT]  Joe Magarac: That would require a protocol change. I'm proposing something that doesn't.
[11:38 AM PT]  Ubit Umarov: as i said.. this is a lot more than protocol
[11:38 AM PT]  Joe Magarac: So old viewers can interoperate with new servers.
[11:38 AM PT]  Ubit Umarov: they could not no matter what
[11:38 AM PT]  Joe Magarac: Why not?
[11:39 AM PT]  Ubit Umarov: bc all code assumes the current flat model
[11:39 AM PT]  Joe Magarac: Not viewer side.
[11:39 AM PT]  Ubit Umarov: maybe not on yours
[11:39 AM PT]  Joe Magarac: Viewer side supports a tree. Even in the C   code.
[11:39 AM PT]  Joe Magarac: When you have an avatar with attachments sitting, you have a 3 level prim hierarchy.
[11:40 AM PT]  Ubit Umarov: yes but avatars are not prims
[11:40 AM PT]  Joe Magarac: They are single objects for object update purposes.
[11:40 AM PT]  Ubit Umarov: the attachments go into fixed predefined attach points etc
[11:40 AM PT]  Ubit Umarov: otehr model
[11:41 AM PT]  Joe Magarac: Yes, avatars and attachments get some special handling, but the tree structrure at the object update level is quire real.
[11:41 AM PT]  Ubit Umarov: ll answer to your proposal is same they give since they decided for the "flat" model
[11:42 AM PT]  Ubit Umarov: :)
[11:42 AM PT]  Joe Magarac: It's been discussed extensively at Creator User Group.
[11:42 AM PT]  Joe Magarac: Several LL people like the idea.
[11:42 AM PT]  Joe Magarac: Off-center pivot points were being discussed, and this looked easier.
[11:43 AM PT]  Joe Magarac: As well as being more general and understandable.
[11:43 AM PT]  Tess JL: Maybe but if I understand right, glTF mesh will require hierarchical linking so they may have to implement it one way or other anyway
[11:43 AM PT]  Joe Magarac: Right.
[11:43 AM PT]  Vincent.Sylvester The same thing happened with linkset data and then it took over ten years for them to consider it only to implement it poorly...
[11:44 AM PT]  Joe Magarac: Things are picking up on the SL side. They hired some game devs, and seem to be overcoming their old "Can't do" mindset.
[11:44 AM PT]  Ubit Umarov: you dont add hierachy withou also adding types of joints for physical etc
[11:45 AM PT]  Joe Magarac: There's quite a bit of new stuff in the SL pipeline. glTF textures, glTF meshes, glTF terrain, non-repeating terrain...
[11:45 AM PT]  Ubit Umarov: when will they release the materials?
[11:45 AM PT]  Tess JL: "Soon" ;-)
[11:45 AM PT]  Joe Magarac: glTF materials is in test now on the SL beta grid.
[11:46 AM PT]  Ubit Umarov: and that twisted heavy use of json gltf ?
[11:46 AM PT]  Joe Magarac: There's a viewer that understands it.
[11:46 AM PT]  Ubit Umarov: i know that even the viewer in suposed to be in QA
[11:46 AM PT]  Ubit Umarov: version 7
[11:47 AM PT]  Joe Magarac: glTF materials are close. They even include mirrors.
[11:47 AM PT]  Ubit Umarov: you call that mirrors?
[11:47 AM PT]  Ubit Umarov: lol
[11:47 AM PT]  Tess JL: But Joe has an important point. LL's dev department is much more serious now than they were during the Six Dark Years. They've been slowly but steadily improving ever since Oz and Ebbe took voer in 2014 and although they still have a long way to go, don't expect them to do as poorly as they did during the 2008-2013 period
[11:48 AM PT]  Joe Magarac: Huge improvement in the last two years in the SL dev team viewer side.
[11:48 AM PT]  Cuga.Rajal lets hope that FS-OS can keep up
[11:48 AM PT]  Joe Magarac: Right.
[11:48 AM PT]  Ubit Umarov: your viewer does those pbr?
[11:49 AM PT]  Joe Magarac: I do not do SL glTF materials yet. But since the underlying model I uses is PBR with the same layers SL is using from glTf, it's just a format decode problem. The renderere is ready now.
[11:49 AM PT]  Joe Magarac: Except for reflections.
[11:49 AM PT]  Tess JL: Ideally we shouldn't keep up, Cuga, we should be ahead of them. ;-)
[11:50 AM PT]  Cuga.Rajal correct :)
[11:50 AM PT]  Andrew Hellershanks nods
[11:50 AM PT]  Cuga.Rajal but it's not "we" it's the FS OS dev team
[11:50 AM PT]  Cuga.Rajal well, for client side
[11:53 AM PT]  Tess JL: glTF PBR isn't actually that different from SL/OS materials. It uses metalicness roghnbess maps rather than RGB specular maps and that may take a bit getting used to but the result isn't that different. The *big* change though, is the addition of reflection probes
[11:53 AM PT]  Cuga.Rajal sorry I came in late, is this a discussion on how to handle the new materials on server side?
[11:53 AM PT]  Ubit Umarov: its collada in jsnn format :p
[11:53 AM PT]  Ubit Umarov: in json
[11:54 AM PT]  Ubit Umarov: and already incomplete..  whatever
[11:54 AM PT]  Ubit Umarov: no..  ppl just telling some changes ll is considering
[11:54 AM PT]  Ubit Umarov: if we wil follow or not,, will depend..
[11:55 AM PT]  Andrew Hellershanks: Hello, Cuga.
[11:55 AM PT]  Lyr Lobo: Hello *smiles*
[11:56 AM PT]  Andrew Hellershanks: Just five minutes left in the hour. A bit too late today if we wanted to pick up the discussion from last week about improved library assets.
[11:56 AM PT]  Tess JL: But the thing about glTF, Ubit, is that it isn't about SL as such, it's about implementing a new official standard for virtual realities
[11:57 AM PT]  Ubit Umarov: yeap several using it
[11:57 AM PT]  Andrew Hellershanks: Are there any other questions/comments for this week?
[11:57 AM PT]  Ubit Umarov: lets see how long
[11:57 AM PT]  Tess JL: Nothing urgent, Andrew
[11:58 AM PT]  Cuga.Rajal nothing urgent or short :)
[11:58 AM PT]  Vincent.Sylvester There is always IRC Tess :)
[11:58 AM PT]  Andrew Hellershanks smiles at Cuga.
[11:58 AM PT]  Andrew Hellershanks: IRC and the mailing lists.
[11:59 AM PT]  Vincent.Sylvester I got something, if someone has some knowledge of the .anim used internally drop me an IM I need some help with that stuff
[12:00 PM PT]  Cuga.Rajal what kind of knowedge?
[12:00 PM PT]  Ubit Umarov: well as you guess the crap ll is doing to add pbr materials does not impress me
[12:01 PM PT]  Joe Magarac: I haven't gotten to .anim yet. Sorry.
[12:01 PM PT]  Cuga.Rajal I like the idea of making a mirror. Thats about the only new use I can think of for me
[12:01 PM PT]  Ubit Umarov: had you ever seen such mirrors?
[12:01 PM PT]  Motoko.Karu mirrors would be useful for some of us... 
[12:01 PM PT]  Ubit Umarov: to call that a mirror.. well
[12:02 PM PT]  Tess JL: This is what LL says about the "PBR mirror": Reflection probes are not intended for use with planar mirrors, and will look incorrect when being used to do so. Planar mirrors have been flagged for future work, but are not directly in the scope of the PBR project.
[12:02 PM PT]  Ubit Umarov: and yeah those will help to keep your room warm with cpu and gpu heat ;)
[12:02 PM PT]  Andrew Hellershanks: :)
[12:03 PM PT]  Joe Magarac: Right.
[12:03 PM PT]  Andrew Hellershanks: Last call for any other items for today.
[12:03 PM PT]  Andrew Hellershanks says thinking we might get done close to the hour for a change.
[12:04 PM PT]  Joe Magarac: Thanks all.
[12:04 PM PT]  Andrew Hellershanks: I see no signs of typing in progress so I will take that as a sign we are done for today.
[12:04 PM PT]  Joe Magarac: Bye for now.
[12:04 PM PT]  Andrew Hellershanks: Thank you all for coming. See you again next week.
[12:04 PM PT]  Andrew Hellershanks: Bye, Joe.
Personal tools
About This Wiki