Chat log from the meeting on 2015-11-17

From OpenSimulator

Jump to: navigation, search
[11:00] Tom.Frost so, last week i subscribed to the opensim-dev list, now i regret i did :)
[11:00] Richardus.Raymaker you can always unsubscribe
[11:00] Richardus.Raymaker it's a bit noisy at this moment
[11:01] Tom.Frost hehe yeah, i'll see where it goes.. but this is one unhealthy community by the looks of it!
[11:01] Aine.Caoimhe the last couple of weeks have been non-typical ones for the mailing list
[11:01] Tom.Frost that's good to know
[11:01] Aine.Caoimhe but perhaps a conversation that is long overdue and has been brewing for quite a while
[11:01] OtakuMegane Desu: Never knew people were so opinionated by an fps readout. :P
[11:01] Tom.Frost yeah, the fps thing was a symbol mostly, i had the impression
[11:01] Shy Robbiani: too much drama... but it has not always been that way
[11:02] Aine.Caoimhe I actually think it's an underlying issue that is causing the most angst
[11:02] OtakuMegane Desu: Probably
[11:02] Aine.Caoimhe it just happened to be this one instance to brought to to light
[11:02] Aine.Caoimhe it could easily have been something else
[11:02] Aine.Caoimhe the proverbial "straw that broke the camel's back"
[11:03] Richardus.Raymaker And what underlying one must eb there. the only one i see is that the are afraid opensim get better and the need to change / relearn things
[11:03] Andrew Hellershanks waves hello
[11:03] Shy Robbiani: sometimes such things can be corrective and healthy
[11:03] Richardus.Raymaker hi andrew
[11:03] Shez Oyen: Hi Robert :)
[11:03] OtakuMegane Desu: It's a guess but at least part is the issue of keeping things like they are/SL-compatible/etc and the ones who want to push on ahead.
[11:03] Aine.Caoimhe yes, that's part of it
[11:03] Richardus.Raymaker Only wish we could dump dae and use fbx standards
[11:03] Shy Robbiani: hi Robert and Andrew
[11:03] Andrew Hellershanks: What is the topic under discussion?
[11:03] OtakuMegane Desu: Drama
[11:04] Tom.Frost :)
[11:04] Shez Oyen: None yet
[11:04] Andrew Hellershanks: :)
[11:04] Aine.Caoimhe also a perceived double-standard between the "in group" and the "not in group" of devs
[11:04] Richardus.Raymaker hi shez. good decription otaku
[11:04] Aine.Caoimhe and I guess the nature of the beast in terms of roadmap (or lack thereof)
[11:04] Shez Oyen: Rich :)
[11:04] Tom.Frost yeah, that does worry me a bit, the lack of a roadmap (and the fact that melanie dismissed anything smelling of roadmap)
[11:05] Tom.Frost but well.. i'll tag along for a bit, see if i can get to grips with the codebase..
[11:05] Tom.Frost it's not as if i have anything to say about it :)
[11:05] Aine.Caoimhe there is also a strong dissimilarity in the way git is used by Opensim devs vs the way most development environments use it
[11:05] Aine.Caoimhe which tends to lead to misunderstandings
[11:05] Tom.Frost btw, is there a way to get rid of the "Unable to fetch profile data at this time." message that keeps popping up in osgrid?
[11:05] Shez Oyen: Neb :)
[11:05] Robert Adams: hello all
[11:06] Nebadon Izumi: hey Robert :)
[11:06] Tom.Frost well, as long as the way git is used in the project is documented, that shouldn't be a problem..
[11:06] Nebadon Izumi: not at all
[11:06] Nebadon Izumi: we dont use git really any differently
[11:06] Richardus.Raymaker ah neb ariived. hi
[11:07] Tom.Frost well, i tend to prefer rebasing over merging.. but that is just because merges can lead to additional changes
[11:07] Nebadon Izumi: its all a matter of preference how it is used
[11:07] Simulator Version v0.5 ruft: OpenSim Dev        dd9b06e: 2015-10-29 22:14:11 +0000 (Unix/Mono)
[11:07] Nebadon Izumi: but we dont do anything unusual with git
[11:07] Aine.Caoimhe I'm sure we don't want to get into that debate's used the way it's used
[11:07] Tom.Frost exactly..
[11:08] Aine.Caoimhe I do have  question about a commit I saw last night though
[11:08] Shy Robbiani: I think the process is not bad
[11:08] Tom.Frost and so far, i haven't seen anything weird in how git is used, for wht its worth
[11:08] Nebadon Izumi: ok well I guess I dont understand is all i wasnt trying to argue
[11:08] Aine.Caoimhe I notice that one of Ubit's "fixes" related to oar import of sit targets
[11:08] Nebadon Izumi: but i cant really think of anything we do that would be not normal
[11:08] Tom.Frost before we move on to aines commit, is itdocumented somewhere?
[11:08] Tom.Frost (the git process)
[11:08] Robert Adams: is this sim running the latest master? is this avinationmerge?
[11:08] Nebadon Izumi: sure
[11:08] Nebadon Izumi:
[11:08] Aine.Caoimhe and apparently the incoming avination merge branch does NOT respect stored sit targets in prims
[11:09] Nebadon Izumi: I thought i saw him maybe address that Aine
[11:09] OtakuMegane Desu: Is there any stability issues with the master now that the merge has happened? I got an instance up and running but after a time it just crashes, no errors or anything.
[11:09] Aine.Caoimhe I saw him disable the test
[11:09] Nebadon Izumi: Ubit is on his way
[11:09] Nebadon Izumi: i just pinged him
[11:09] Aine.Caoimhe but I didn't see anything yet to add support for loading oar with sit targets stored
[11:09] Nebadon Izumi: he like myself was distracted hehe
[11:09] Nebadon Izumi: ok see what he says
[11:09] Aine.Caoimhe but perhaps I misread or misunderstood the meaning of his push
[11:10] Nebadon Izumi: he is wrestling unit tests at the moment
[11:10] Jak Daniels: he also added a LegacySitOffsets to the .ini
[11:10] Nebadon Izumi: its possible its on the to-do list
[11:10] Richardus.Raymaker did sit targets got saved before ?
[11:10] Jak Daniels: to make that still happen I think
[11:10] Aine.Caoimhe nods
[11:10] Tom.Frost neb: from a quick glance that's mostly just a git manual, right?
[11:10] Aine.Caoimhe I just wanted to make sure that the "plan" wasn't do make it impossible to store and reload sit targets
[11:10] Nebadon Izumi: well it shows there how you can make patches and provide them, how to use git
[11:10] Nebadon Izumi: we can pull code from github
[11:10] Nebadon Izumi: though a lot of times its not always easy
[11:11] Nebadon Izumi: merging code is complicated
[11:11] Andrew Hellershanks: Sit offsets changed a few times. Which version is considered legacy?
[11:11] Jak Daniels: before avn merge I guess
[11:11] Ubit Umarov: ( why whould we have a plan to make it impossible to store sit targets? )
[11:11] Nebadon Izumi: the big problem is most people wait to long and change a lot of things
[11:11] Nebadon Izumi: merging that is complicated and time consuming
[11:11] Shy Robbiani: Tom, the correct way is to make patches and add them to the Mantis...
[11:11] Nebadon Izumi: like what Ubit is going through
[11:11] Tom.Frost nebadon: agreed
[11:11] Nebadon Izumi: hehe
[11:11] Aine.Caoimhe I'm hoping you didn't
[11:11] Nebadon Izumi: its painful
[11:11] Tom.Frost shy: that's perfectly normal then :)
[11:12] Nebadon Izumi: I would say the easier more reliable way is patches
[11:12] Aine.Caoimhe I just read your git about disabling loading them from OAR and became a little concerned, that's all
[11:12] Nebadon Izumi: however if you can gain the trust of a core dev
[11:12] Nebadon Izumi: we have no problem cherry picking from other repositories
[11:12] Nebadon Izumi: if the code is clean
[11:12] Nebadon Izumi: no one is just going to do it out of the blue though
[11:12] Andrew Hellershanks: Um... disable loading of sit settings??
[11:13] Aine.Caoimhe
[11:13] Ubit Umarov: what happened is that avination has diferent math to get sits closer to SL. and it was merged, we hoped changes whould be minor.. they where not ..
[11:13] Andrew Hellershanks: While reviewing the AVN merge changelog I cherry picked a couple of changes for a grid I use.
[11:14] Aine.Caoimhe if a prim has no pre-existing sit target your new stuff will be amazing
[11:14] Shy Robbiani: are there any changes with LSL?
[11:14] Aine.Caoimhe but there are occasions where a specific sit target needs to be set
[11:14] Aine.Caoimhe and it would be a major annoyance to have to do that via script on every single oar load and/or sim reset
[11:15] Nebadon Izumi: its nice to even have not scripted sit working at all
[11:15] Ubit Umarov: yeap thats why it was changed
[11:15] Nebadon Izumi: before this is never really worked right that I recall
[11:15] Shy Robbiani: that's true but we had our workarounds
[11:15] Ubit Umarov: ( just added a option so not to loose the new math, that can be used )
[11:15] Aine.Caoimhe if you set a sit target ina prim i towkr ed perfectly
[11:15] Nebadon Izumi: had to really orient yourself correctly to get even 1/2 decent results
[11:15] Aine.Caoimhe even without a script in it
[11:15] Andrew Hellershanks: hmm.... If there is a known (or way to calculate the) offset, another option is a command to adjust the sit positions.
[11:16] Aine.Caoimhe but I agree that the new way of handling a prim without a sit target is great
[11:16] Ubit Umarov: see SL wike Andrew.. its a mess
[11:16] Ubit Umarov: wiki
[11:16] Aine.Caoimhe the new method should occur when the prim has an existing sit target of Zero_VECTOR and ZERO_ROTATION
[11:16] Richardus.Raymaker the wiki got not better since the changed some things this year
[11:17] Aine.Caoimhe else it should respect the stored values
[11:17] Aine.Caoimhe (and be able to store and reload them via OAR)
[11:17] Ubit Umarov: aine.Caoimhe
[11:17] Ubit Umarov: this as nothing to do with stored values
[11:18] Ubit Umarov: it is a offset added on top of sit target
[11:18] Ubit Umarov: a old SL BUG
[11:18] Aine.Caoimhe the 0.4m z-axis one?
[11:18] Ubit Umarov: that is a bit more complex that just 0.4
[11:19] Aine.Caoimhe ah..the one using the volume calculation thingy
[11:19] Ubit Umarov: depends on avatar size etc
[11:19] Aine.Caoimhe right
[11:19] Shy Robbiani: ok
[11:19] Aine.Caoimhe okies
[11:19] Ubit Umarov: at avination we just changed the math a bit
[11:19] Ubit Umarov: also on some LSL fundtions related to sits
[11:19] Aine.Caoimhe I just wanted to be sure since the way that commit says it made me fear that sit targets were going to be ignored
[11:20] Aine.Caoimhe or at least not stored
[11:20] Ubit Umarov: nopes its math on top of them
[11:20] Aine.Caoimhe hence why I asked
[11:20] Shy Robbiani: me too
[11:20] Aine.Caoimhe kk...great!
[11:20] Andrew Hellershanks: Aine, it still sounds like sit positions won't be just what is stored in the object.
[11:21] Ubit Umarov: now the new math did cause visible diference on all this sits here
[11:21] Nebadon Izumi: I guess everyone is also fully aware that AVN merge code is now in master
[11:21] Ubit Umarov: so..  another OPTION  lol
[11:21] Nebadon Izumi: so no need to test seperate branch for those who are
[11:21] Nebadon Izumi: just switch back to master now
[11:21] Aine.Caoimhe yes, I'll still have to go and redo all of my existing stripts that calculate those offsets
[11:21] Tom.Frost yep
[11:21] Nebadon Izumi: if you watch irc or jenkins ignore the fails for now
[11:21] Aine.Caoimhe but it's nothing new for changes to require script rewrites
[11:21] Tom.Frost oh this is odd.. as soon as i start typing, my sit position changes
[11:21] Tom.Frost (i am raised up)
[11:22] Nebadon Izumi: code is failing on unit tests
[11:22] Nebadon Izumi: becasue of new code
[11:22] Andrew Hellershanks: Nebadon, I read that the other day and was expecting to see a big "Merge from avinationmerge branch" message in the git log.
[11:22] Nebadon Izumi: Ubit is sadly very occupied with this at the moment
[11:22] Nebadon Izumi: you can see the log on the avinationmerge branch
[11:22] Aine.Caoimhe it has been of yesterday
[11:22] Nebadon Izumi: i dont think the merge itself will spit out a 1000 page long report :)
[11:22] Nebadon Izumi: there is a ton of changes
[11:23] Nebadon Izumi: and new stuff
[11:23] Andrew Hellershanks: I have the original merge changelog summary down to about 2500 lines.
[11:23] Ubit Umarov: nopes my fixes to make jenkins happy will ;)
[11:23] Nebadon Izumi: i would suggest anyone who is doing a > git checkout master
[11:23] Nebadon Izumi: also do a git reset --hard and git clean -d -f
[11:23] Nebadon Izumi: after switching back to master
[11:23] Andrew Hellershanks: Can't do that yet. I would lose some work.
[11:23] Nebadon Izumi: or just start a fresh clone, probably more sane
[11:23] Nebadon Izumi: well yes if you have code checked in
[11:24] Nebadon Izumi: imjust talking about testers
[11:24] Andrew Hellershanks: I have some stashed changes that are causing a merge conflict.
[11:24] Shy Robbiani: how works jenkins? I was always wondering where these messages come from in the IRC
[11:24] Nebadon Izumi: yes those are never fun
[11:24] Nebadon Izumi:
[11:24] Nebadon Izumi: he is our code testing butler :)
[11:24] Andrew Hellershanks: I know. I'm going to try saving the changes, reset my tree, then apply the changes.
[11:24] Nebadon Izumi: everytime someone checks in code, Jenkins compiles and runs unit tests
[11:24] Nebadon Izumi: to make sure new code isnt breaking expected behaviors we test for
[11:24] Shy Robbiani: can we all use it?
[11:25] Aine.Caoimhe how sure are we that nothing in the new stuff will have any adverse affect on a grid's Robust/assets  if someone is using it in a region in that grid and the grid hasn't also updated to that same master?
[11:25] Tom.Frost nebadon: that's a good thing!
[11:25] Nebadon Izumi: you can read its reports
[11:25] Nebadon Izumi: not much to use about it
[11:25] Nebadon Izumi: its quite auotmated
[11:25] Nebadon Izumi: if you sit in opensim-dev channel on freenode
[11:25] Nebadon Izumi: you can see it happen
[11:25] Shy Robbiani: will all code go through?
[11:25] Nebadon Izumi: anything that checks into out git repo master will
[11:25] Nebadon Izumi: our*
[11:26] OtakuMegane Desu: You can watch in IRC, it usually pops up a couple minutes after a commit
[11:26] Nebadon Izumi: its been failing unit testing for about 48 hours now, it looks bad but its really not
[11:26] Ubit Umarov: (Aine only testing will give a answer to that )
[11:26] Aine.Caoimhe ie you're sure that if grid is running 0.8.2 and a region tries to run 0.8.3dev there won't be any explosions or other fireworks at grid level
[11:26] Aine.Caoimhe kk
[11:26] Shy Robbiani: does it mean there is no need to run all the tests myself prior to commit patches?
[11:27] Tom.Frost shy: ideally jenkins is a safety net
[11:27] Nebadon Izumi: its always good to be safe in your environments, just because jenkins thinks its right, its not always correct
[11:27] Tom.Frost not an excuse to not make sure your code compiles :)
[11:27] Nebadon Izumi: yea Jenkins gets the low hanging fruit
[11:27] Nebadon Izumi: basically
[11:27] Aine.Caoimhe I don't mind wiping my own system and having to revert
[11:27] Nebadon Izumi: all of us need do top level testing
[11:27] OtakuMegane Desu: Unit tests test pre-selected things. There's a lot of stuff that can break and it'll never know.
[11:27] Aine.Caoimhe but I don't want to risk killing the grid doing it
[11:27] Shy Robbiani: :)
[11:28] Ubit Umarov: there is another issue that may cause some concern..  about physics meshs
[11:28] Nebadon Izumi: it seems unlikely you would kill the grid, we are testing it here on osgrid
[11:28] Nebadon Izumi: its been great so far
[11:28] Andrew Hellershanks: bbiab
[11:28] Aine.Caoimhe I wouldn't expect it to
[11:28] Nebadon Izumi: kk Andrew
[11:28] OtakuMegane Desu: Grid-killing stuff is very rare and usually gets caught almost immediately if something does slip in.
[11:28] Aine.Caoimhe but I also wouldn't want to find out the hard way
[11:29] Shy Robbiani: the only concern I asked in the mailing list... will existing meshes still work?
[11:29] Nebadon Izumi: ya always best practice to backup before any big changes like this of course
[11:29] Ubit Umarov: ( the ones upload without any higher resolution setting for physics mesh )
[11:29] Nebadon Izumi: so far ive not noticed anything not work myself
[11:29] Nebadon Izumi: Sisyphus is heavily laden with mesh
[11:29] Nebadon Izumi: no console errors, no iin world strangeness
[11:29] Ubit Umarov: some will not work as before on ubOde
[11:29] Ubit Umarov: and in future on all
[11:29] Nebadon Izumi: yea im testing Bulletsim still
[11:29] Shy Robbiani: ok... I have lots of mesh I can use for tests
[11:29] Nebadon Izumi: i have not tested ubode yet as a disclaimer
[11:30] Aine.Caoimhe Bullet is being removed?
[11:30] Nebadon Izumi: noo
[11:30] Ubit Umarov: nopes
[11:30] Nebadon Izumi: original ODE may down the road
[11:30] Nebadon Izumi: if no one gripes
[11:30] Aine.Caoimhe looks at Robert having a heart attack :p
[11:30] Ubit Umarov: lol
[11:30] Nebadon Izumi: but really it not a huge deail to have both either
[11:30] Aine.Caoimhe not at all
[11:30] Nebadon Izumi: or even make original ode an external module on forge
[11:31] Aine.Caoimhe it  was Ubit's "in future on all" comment that made me wonder
[11:31] Jak Daniels: Ubit is this the issue with bullet using the high lod of the object as physics if no physics mesh is present whereas ubode uses convex hull?
[11:31] Ubit Umarov: yeap jak
[11:32] Robert Adams: If an object is made physical, BulletSim uses and algorightm to generate a set of conex hulls from the mesh
[11:32] Ubit Umarov: current physics ( old ode also )  if doesn't find a mesh,  uses a visual one
[11:32] Robert Adams: if there is no physical mesh supplied in the mesh
[11:32] Jak Daniels: couldn't ubode do the same thing and let you choose afterwards when editing the prim?
[11:32] Jak Daniels: but create the mesh using high lod if its not there
[11:33] Robert Adams: BulletSim does the same -- for static objects, it uses the display mesh as the collision mesh if one is not supplied
[11:33] Richardus.Raymaker And that works pretty good with mesh in opensim
[11:33] Ubit Umarov: Robert  Mesmerizer does give you a visual LOD mesh
[11:33] Ubit Umarov: that mix with visual causes issues with the use of Physics shape type etc
[11:34] Robert Adams: BUlletsim reduced the LOD to create the mesh, but, yes, it uses the Mesherizer to make a collision mesh if one is not supplied
[11:34] Robert Adams: it will use the HACD algorithm to create a set of convex hulls if the object is made physical
[11:34] Richardus.Raymaker Am only not sure if it's the viewer or opensim that go weird when you press by accident the analyze button in the uploaded. i know never press that button.
[11:35] Nebadon Izumi: ya never push that button
[11:35] Tom.Frost oh..
[11:35] Richardus.Raymaker yuck V-HACD. bad experience with that
[11:35] Ubit Umarov: well ubOde never does that mix. If onlt the convex is present the thing is convex
[11:35] Tom.Frost flüstert: how are people supposed to know they should not push that button? :)
[11:35] Robert Adams: BulletSim doesn't use the lowest LOD setting in Meshmerizer as it (used to?) makes meshing mistakes for prim openenings if sett too low
[11:35] Richardus.Raymaker flüstert: Tom " stick a label on it "Do not press" :)
[11:35] Aine.Caoimhe no, don't do that
[11:36] Aine.Caoimhe it's like read me notecards....people do the opposite
[11:36] Nebadon Izumi: use a sharpie to put an x over the button on your screen :P
[11:36] Richardus.Raymaker i think aine did bite the bait :)
[11:36] Tom.Frost i didn't know i wasn't supposed to click that button.. i think that goes for a lot of people :)
[11:36] Robert Adams: Simple prims (convex shapes like cylinders without twists or holes) are not turned into hulls but are used as single convex shapes
[11:37] Nebadon Izumi: ya its unfortunate those cant be turned off for opensim, i beleive that analyze stuff in SL is all done server side, or pehraps with stuff thats in the stock SL viewer that TPV are not allowed to have
[11:37] Nebadon Izumi: probably server side though
[11:37] Jak Daniels: I like being able to upload a mesh without physics if I want the physics mesh to be the same as the high lod. Saves heaps on upload times. You can still change it to a hull afterwards
[11:37] Aine.Caoimhe yes....disabling that for Opensim would be really, really nice
[11:37] Aine.Caoimhe since it tends to hang on it too
[11:37] Richardus.Raymaker Jak, uses lowest physics setting is best then
[11:38] Ubit Umarov: setting physics to a low LOD  should cost nothing
[11:38] Andrew Hellershanks is back and caught up on the discussions
[11:38] Jak Daniels: I use it like that for mesh terrain
[11:38] Ubit Umarov: that will fail
[11:39] Jak Daniels: on ubode yes but not bullet
[11:39] Ubit Umarov: for terrain select the HIGH lod
[11:39] Nebadon Izumi: if you truly want no physics on your mesh
[11:39] Ubit Umarov: it will fail on bullet also, most likelly
[11:39] Nebadon Izumi: make a collada file with 3 vertices a single triangle
[11:39] Nebadon Izumi: and set that as your phyusics
[11:39] Nebadon Izumi: never have it be absolutely nothing
[11:39] Ubit Umarov: if you don't want physics, this is not a issue..
[11:40] Tom.Frost gawd that fresh clone is taking forever.........
[11:41] Andrew Hellershanks: Can someone talk about the changes in prim linking order I saw mentioned in the change log?
[11:41] Ubit Umarov: meshs that will not need to collide, or are ok with the convex hull can be uploaded without selecting phsyics mesh
[11:41] Andrew Hellershanks: It sounds like the order of prims when merging two link sets is going to revert to an older numbering method.
[11:41] Shy Robbiani: I didnt understand that either
[11:42] Ubit Umarov: Andrew i whould if i remembered the details :p
[11:42] Andrew Hellershanks: If I have two links sets with prims number 1234 I want the merge of the two to give me 12345678.
[11:42] Ubit Umarov: think old code did a mess on linking
[11:42] Andrew Hellershanks: The old way was to result in 16782345.
[11:43] Ubit Umarov: when linking 2 linksets etc
[11:43] Andrew Hellershanks: I didn't understand if some code regression snuck in that needed fixing or some other change was happening.
[11:43] Shy Robbiani: hmmm... that's true... I always thought it's the viewer
[11:43] Nebadon Izumi: i remember struggeling with that when i made my watch long while back
[11:43] Nebadon Izumi: i ended up not relying on prim order
[11:44] Shy Robbiani: me too *lol*
[11:44] Andrew Hellershanks: Long ago prim link numbering was bizarre. It got fixed eventually. I know I had complained about it. Now I don't know what's happening after his merge.
[11:44] Andrew Hellershanks: I'll be testing it. I have some objects designed for testing prim link numbering.
[11:44] Aine.Caoimhe even more interestingly, the link number reported by the viewer in some linksets is not correct
[11:44] Nebadon Izumi: ya have to test :)
[11:44] Ubit Umarov: well we did work on that some 3 yrs ago.. don't remember the details now
[11:44] Andrew Hellershanks: Aine, yea that is also odd.
[11:44] Richardus.Raymaker yes, you betetr made the script so it finds the link number first
[11:45] Aine.Caoimhe to do it reliably you have to name each link and then do a hard check of it via script
[11:45] Richardus.Raymaker yup
[11:45] Andrew Hellershanks: Richardus, I have objects with 255 prims that need to be in a given order. I'm not going to sit and number all of them.
[11:45] Aine.Caoimhe which is a PITA if it's a large linkset
[11:45] Shy Robbiani: on Metropolis I have a test set too that I can bring over
[11:45] Richardus.Raymaker yes aandrew. it's a bug
[11:46] Andrew Hellershanks: Importing link sets from an XML file via a viewer never gets you the original prim link order. That's a different issue.
[11:46] Ubit Umarov: well guess testing of this code will now have a boost
[11:46] Nebadon Izumi: i hate taht collada uploads are not consistent either
[11:46] Nebadon Izumi: its almost random
[11:46] Andrew Hellershanks: I'll test master once I can solve my merge isue.
[11:46] Ubit Umarov: so far only a few did testing
[11:46] Aine.Caoimhe yes
[11:47] Ubit Umarov: so sure lots of issues will come up
[11:47] Nebadon Izumi: if i upload 30 piece collada, every time the root prim is different
[11:47] Aine.Caoimhe or at least if dae imports are in any way predictable I haven't yet uncovered the method
[11:47] Nebadon Izumi: it works with Radegast
[11:47] Nebadon Izumi: 100% consitent every time
[11:47] Aine.Caoimhe interesting
[11:47] Andrew Hellershanks: nebadon, I just know the root prim is not the same as it was when I saved the object. Don't know if that is a viewer issue or not.
[11:47] Aine.Caoimhe so that points to it being a viewer code thing
[11:47] Nebadon Izumi: yea seems to be
[11:48] Nebadon Izumi: radegast of course has other problems with the mesh upload
[11:48] Andrew Hellershanks: nebadon, Does using Radegast result in the same root prim as the original exported object?
[11:48] Nebadon Izumi: its great for terrain or anything with full detail physics
[11:48] Nebadon Izumi: that doesnt need LOD
[11:48] Aine.Caoimhe maybe analizing it is done asynch and the order is determined by the order each one finishes
[11:48] Nebadon Izumi: everything else its not good
[11:48] Nebadon Izumi: Andrew I cant say for sure
[11:48] Andrew Hellershanks: k
[11:48] Nebadon Izumi: i know with the terrain our custom software generates
[11:48] Nebadon Izumi: its 100% consistent the SW tile every time
[11:49] Nebadon Izumi: exactly what i need it to be
[11:49] Andrew Hellershanks nods
[11:49] Jak Daniels: sampler and surface is not supported, but I have a fixed mesh upload cmdline libomv client on my github
[11:49] Nebadon Izumi: but i upload same tiles with viewer and its collada roullete
[11:49] Aine.Caoimhe I also frequently run into it hanging or taking insanely long times to analyse a linkset whereas doing them one by one is almost instant
[11:49] Andrew Hellershanks: I haven't figured out how to import a saved object in Firestorm.
[11:49] Nebadon Izumi: yes i have all kinds of wierd glitchy behavior uploading models
[11:49] Nebadon Izumi: unexplainable things
[11:50] Nebadon Izumi: and the stuff always just instantly uploads on radegast
[11:50] Aine.Caoimhe which is why I really with it would treat mesh position as its root instead of its bounding box center
[11:50] Nebadon Izumi: a mesh upload on radegast takes like 10 seconds
[11:50] Nebadon Izumi flüstert: lol
[11:50] Nebadon Izumi: but they are heavy
[11:50] Aine.Caoimhe (ie root as in "Blender location")
[11:50] Nebadon Izumi: which sucks
[11:50] Jak Daniels: i can script upload mesh tiles using my cmdline version neb... saves even more thime ;)
[11:50] Richardus.Raymaker yeah, mesh upload can get stuck after pressing the button bottom left
[11:51] Richardus.Raymaker calculate can be hanging
[11:51] Aine.Caoimhe it would also make non-physical rotations a lot easier to handle
[11:51] Aine.Caoimhe with llTargetAlpha() on a 3-bladed propeller you have to make invisible mesh portions to make the bounding box center match the center of rotation
[11:52] Aine.Caoimhe er...llTargetOmega() I mean
[11:52] Aine.Caoimhe or only ever use even-numbered prop blade
[11:52] Ubit Umarov: aine.Caoimhe
[11:53] Andrew Hellershanks: grr... my cat is calling for me again. :P
[11:53] Ubit Umarov: you need to make things like that all over
[11:53] Aine.Caoimhe sure
[11:53] Aine.Caoimhe I'm just saying it would be *nice* not to have to
[11:53] Andrew Hellershanks: Richardus, I don't know what the calculation does but it can take a long time depending on the mesh.
[11:53] Ubit Umarov: ubOde is diferent on that..  but will not claim its ok ;:)
[11:54] Aine.Caoimhe it would also make it a lot easier to upload larger linksets and get them all positioned correctly
[11:54] Richardus.Raymaker Last time it where hanging i could fix it be relogging or restarting the simulator and not press analyze
[11:54] Richardus.Raymaker But still would love to use FBX
[11:55] Aine.Caoimhe the fun thing with it hanging on uploads is you can retry the same model multi times and sometimes it will work perfectly and almost instantly...other times it will take half an hour...other times it will never finish
[11:55] Aine.Caoimhe and using the identical dae each time
[11:56] Richardus.Raymaker yup. something like that
[11:56] Aine.Caoimhe that's a viewer issue though....I've had a FS Jira in on that which just celebrated its 2nd birthday
[11:57] Aine.Caoimhe I was going to bake it a mesh cake but it wouldn't upload either
[11:59] Tom.Frost lol
Personal tools
About This Wiki