Chat log from the meeting on 2018-07-10

From OpenSimulator

Jump to: navigation, search

- Announcement I: the branch httptests is merged into master again, after the dev-machine was upgraded by Andrew
- The new development-branch is now officially 0.9.1
- The new development code needs at least .NET 4.6 on Windows and Mono 5.12 on Linux
- Macs have issues with 64-bit builds - just build as 32-bit for now until the issue gets fixed
- SSL/TLS still not fully implemented
- Announcement II: Search for regions via map now searches for names that contain the search-string - no need for us to remember the beginning of the name anymore. The viewer have to apply a patch though, so it will take to one of the next releases for search to work that way...
- Discussion of FSEnforceStrictObjectCheck and malformed meshes and prims

[11:10] Andrew.Hellershanks I think we can get started now.
[11:10] Andrew.Hellershanks We have one bit of news this week for those who don't follow the mailing list.
[11:12] Andrew.Hellershanks I installed a new mono on the machine and this meant that Ubit was able to merge the code in the httptests branch back in to master.
[11:12] Andrew.Hellershanks For anyone who wasn't happy about master not being where the main code was being developed they can rest easy.
[11:13] Bill.Blight always shocked me how many people were hung up on the names, master or httptests
[11:13] Andrew.Hellershanks nods to Bill.
[11:14] LaNani Sundara: that closes the door on the release version Up to 1.0 ?
[11:14] Andrew.Hellershanks It could be that some people may not have known for sure which branch had the latest code on it. Now they won't have that problem.
[11:14] LaNani Sundara: its such a big code change now its merged, it deserves a new version number :)
[11:15] Ubit Umarov: as i said on a email, there may still be some 0.9.0.x bug fixes releases.. if fixes are simple etc
[11:15] Bill.Blight Personally I think it should at least be .9.2.0
[11:15] Ubit Umarov: it has a new version number 0.9.1 :p
[11:16] Bill.Blight Problem is, there were people out there already calling their versions 0.9.1 ... like , ohhh, ummm , osgrid
[11:16] LaNani Sundara: the previous master was known as 0.9.1 to many until you made it
[11:16] Andrew.Hellershanks Bill, I seem to remember a similar problem in the past where some people dropped 0. or .0 out of a release number.
[11:17] LaNani Sundara: 0.9.2 for the new one would be appropriate
[11:17] Ubit Umarov: it was 0.9.1 Dev.. Dev matters :)
[11:18] Ubit Umarov: all that was on master was basicly fixes to
[11:18] Ubit Umarov: so.. a fixes release...
[11:18] LaNani Sundara: maybe add an ub like you did with ODE ;p
[11:18] Ubit Umarov: no need for a release candidate etc
[11:19] Ubit Umarov: now most the things that arrived from httptests, will need a RC
[11:19] TG.Lucan would be useful to some on later win 10 for the dotnet to be 4.6.2?
[11:19] Ubit Umarov: ( a few may still get into a 0.9.0.x.. but guess not )
[11:20] Ubit Umarov: i did consider 4.6.2.. but well decided for 4.6 for now
[11:21] TG.Lucan people will need to install 4.6 not a clean update
[11:21] LaNani Sundara: had a few people come to me with problems using the last release with older mono versions, advised them to install new mono and problems were gone
[11:21] TG.Lucan another level for things to go wrong
[11:22] Ubit Umarov: well i did consider 4.5 also :P
[11:22] Ubit Umarov: was my first choice ;)
[11:22] Andrew.Hellershanks When code is pushed to master there is a Jenkins process that automatically compiles the changed code using mono 5.12
[11:22] TG.Lucan that would solve many problem :)
[11:23] Ubit Umarov: and actually tested 4.7.1
[11:23] Ubit Umarov: but something told me 4.7.1 could be a bit 2 much ;)
[11:23] TG.Lucan I also tested on 4.7.1 for curious interest
[11:24] Ubit Umarov: and when i did it mono5 was not ready for it
[11:24] TG.Lucan xbuild does not support 4.7.1
[11:25] LaNani Sundara: ... well i guess now i will have to start taking the new master apart and improve upon it  :p (cuts out those nasty throttles ;P )
[11:25] Ubit Umarov: msbuild is the thing to use
[11:25] TG.Lucan sure
[11:26] Ubit Umarov: i did change our site on instructions
[11:26] TG.Lucan another bump for folk on earlier linux
[11:26] Ubit Umarov: those with wiki access may improve it
[11:26] Ubit Umarov: for now i did mention that Mac have issues with 64bit
[11:27] Ubit Umarov: we have a lib giving issues
[11:27] Gavin.Hird the problem is libopenjpeg-dotnet.dylib
[11:27] Andrew.Hellershanks Master can be built with xbuild as I still use that for compiles but it will report that you should be using msbuild.
[11:27] Gavin.Hird it was built 32 bit with GCC that no longer exist on the Mac, and the code is not any friend of clang at all
[11:28] Ubit Umarov: yes mantis 3308
[11:28] Ubit Umarov: we have all hopes on you to help on that Gavin :)
[11:28] Gavin.Hird we'll see what happens
[11:28] Ubit Umarov: thanks :)
[11:29] Gavin.Hird first prio is a viewer that will run at all on the upcoming macOS 10.14
[11:29] Gavin.Hird not a single viewer even makes it to a login screen on the beta
[11:29] Ubit Umarov: btw you said mono -server
[11:29] LaNani Sundara: FireStorm team not on it?
[11:29] Ubit Umarov: on linux -desktop seems a better option
[11:29] Sheera Khan: first signs of the deprecation of openGL?
[11:29] Gavin.Hird he, the LL code crash immesdiately
[11:30] Gavin.Hird and virtually all viewers use it
[11:30] Gavin.Hird it crash in LLSingleton
[11:30] Gavin.Hird no, it has noting to do with OpenGL
[11:31] Gavin.Hird When Apple makea deprecation it usually takes 5-6 years before it goes away
[11:31] Gavin.Hird the only thing that can happen is they release new hardware that does not support OpenGL at all
[11:31] Gavin.Hird all existing kit does
[11:35] LaNani Sundara: so... is https fully supported now?
[11:35] LaNani Sundara: :)
[11:35] Ubit Umarov: no
[11:35] Ubit Umarov: its master now lol
[11:35] LaNani Sundara: i meant ssl
[11:35] Gavin.Hird the viewers don't support it either
[11:36] TG.Lucan we are using mono with MONO_ENABLE_COOP set, improves the gc efficiency, though at this time we do not have specific timings available
[11:36] Ubit Umarov: no ssl is still work todo
[11:36] Leighton.Marjoram even SSL logins would be good
[11:36] Ubit Umarov: or tls
[11:36] Leighton.Marjoram a good start I mean
[11:36] Leighton.Marjoram oops always forget its not SSL any more
[11:36] Ubit Umarov: yeap :)
[11:37] Ubit Umarov: but all call it ssl
[11:37] Gavin.Hird http2 has been added to viewer dev builds, but it's buggy
[11:37] Ubit Umarov: and that animesh thing ?
[11:37] Andrew Hellershanks: If it isn't called SSL anymore what is it called?
[11:37] Gavin.Hird last LL dev meeting said it is complicated so it progress solvly
[11:37] Leighton.Marjoram TLS
[11:37] Sheera.Khan there's a channel in the SL grid supporting animesh already
[11:38] Andrew Hellershanks: oh. I thought TLS was different.
[11:38] Leighton.Marjoram v1.0-1.3
[11:38] Sheera.Khan BlueSteel I think...
[11:38] LaNani Sundara: haha
[11:38] Gavin.Hird plus they also need bakes on mesh to work first
[11:38] Ubit Umarov: ppl still call it ssl andrew but the ssl protocol is now mainly blocked on many uses
[11:38] Leighton.Marjoram TLS 1.3 isnt widely used yet
[11:38] Ubit Umarov: protocol is tls yes
[11:39] Andrew Hellershanks: I think I still call it SSL because of the names of the libraries.
[11:39] LaNani Sundara: :)
[11:39] Ubit Umarov: .net upgrade was also a help on that
[11:39] Leighton.Marjoram SSL is used for marketing mainly thats where the confusion comes from.
[11:39] Andrew Hellershanks: One other trivial bit of news this week affects searching for regions using the map dialog in the viewer.
[11:40] Andrew Hellershanks: Users have complained about not being able to search for a region using part of a region name where what they are searching for isn't the start of the region name.
[11:40] Gavin.Hird I tested it and the result was the same before and after the commit
[11:40] Bill.Blight requires a patched viewer
[11:41] Gavin.Hird what patch is that?
[11:41] Andrew Hellershanks: Its something I've been wanting to look in to for a while. I pushed a change to the OS code but a patch is also required to the viewer in order for it to work fully.
[11:41] Andrew Hellershanks: Its a one line change to both OS and to the viewer.
[11:42] Andrew Hellershanks: I haven't filed bug reports yet for any viewer asking for the change to be implemented in the viewer.
[11:42] Ubit Umarov: think for now the viewer patch is on andrew 's pc :)
[11:42] LaNani Sundara: would viewer discard extra search results then?
[11:43] Bill.Blight hehe and on mine
[11:43] Andrew Hellershanks: If you were looking for regions with "sandbox" in the name the viewer expects you are looking for regions with names starting with "sandbox". If it receives additional results that don't start with sandbox it discards them.
[11:44] LaNani Sundara: .. silly viewer
[11:44] Andrew Hellershanks: The one line patch for the viewer just ensures that the search string exists somewhere in the returned names.
[11:45] Andrew Hellershanks: I don't know why the viewer bothers to even do any sort of validation on the search results but that could just be some strange "feature" inherited from the LL viewer.
[11:45] LaNani Sundara: my thought too Andrew... viewer should just send the query and display the result
[11:46] Andrew Hellershanks: That's what I would have expected.
[11:46] LaNani Sundara: me too
[11:46] Ubit Umarov: you all see my email about Firestorm setting that turns prims invisible ( that info is also available elsewhere )
[11:46] Andrew Hellershanks: Unfortunately that is not the case so all viewers will need a change.
[11:47] LaNani Sundara: yes its been noticed since the new firestorm came out
[11:47] Gavin.Hird what is that setting called again
[11:47] Ubit Umarov: FSenforce something lol
[11:47] LaNani Sundara: i happily stay on FS 5.0.7
[11:47] Ubit Umarov: FSEnforceStrictObjectCheck
[11:48] Gavin.Hird right
[11:48] Andrew Hellershanks: I'm on 5.1.8 as I needed to compile FS to test the change to the viewer code.
[11:48] LaNani Sundara: who asked for strict checks anyway, thats like the viewer discarding search results
[11:48] Gavin.Hird I think FS has it on by default
[11:49] Andrew Hellershanks: I can't test the change in Singularity as there is a single line compile error when using a recent version of GCC.
[11:49] Ubit Umarov: it has Gavin.Hird
[11:49] Ubit Umarov: so some confusion around
[11:49] LaNani Sundara: ,, wish viewer devs would spend their time on more important stuff
[11:49] Andrew Hellershanks: Ubit, Does it affect only mesh prims or all prims?
[11:50] Ubit Umarov: all prims i think
[11:50] Gavin.Hird the description text says prims, but it is primarily mesh
[11:50] Gavin.Hird there was a time when maeprims did not always behave
[11:50] Gavin.Hird megaprims
[11:50] Ubit Umarov: had a prim case on #osgrid
[11:50] Andrew Hellershanks: Yup. It is set to true in 5.1.8
[11:51] Ubit Umarov: a box that puffed while working on it
[11:51] Andrew Hellershanks: FSEnforceStrictObjectCheck: Force malformed prims to be treated as invalid. This setting derenders all malformed prims, even those that might not cause obvious issues. Setting to false will allow bad prims to render but can cause crashes. (Disabled on Ope
[11:51] Andrew Hellershanks: hm... Was it trying to say that the setting/feature is disabled on OpenSim grids?
[11:51] Ubit Umarov: well it is not disabled on opensim it seems
[11:51] Ubit Umarov: ( by default )
[11:52] Gavin.Hird I think it originally was, but then last fall it was changed to on
[11:52] Bill.Blight it is now in their latest dev code
[11:52] Bill.Blight set to false for opensim I do believe
[11:52] Ubit Umarov: well it should a option on opensim
[11:52] Ubit Umarov: so ppl can test prims
[11:52] Bill.Blight just like them forcing the disable of http pipeline in opensim
[11:52] LaNani Sundara: what are "bad" prims
[11:53] Bill.Blight even if you turn it on, it is off
[11:53] Ubit Umarov: not on ..11
[11:53] Gavin.Hird examples of bad prims are some of the original megaprims
[11:53] Ubit Umarov: not all combinations of the parameters are legal lanani
[11:54] Andrew Hellershanks: I've seen messages in instance consoles about bad sculpties. I suspect the setting would affect the display of those types of prims but they usually don't appear anyway if the sculpty data is bad.
[11:54] Ubit Umarov: and viewers do accept some bad still
[11:54] LaNani Sundara: ok
[11:54] Ubit Umarov: and i made a bad one on meshes upload :p
[11:54] LaNani Sundara: shouldnt they then just make their viewers not crash when they accept those
[11:54] Bill.Blight the thing to remember is broken is broken, even if you could use it, broken still means broken
[11:54] Ubit Umarov: ( some meshes only )
[11:55] Andrew Hellershanks: The slightly disturbing comment about the setting is that last part of "This setting derenders all malformed prims, even those that might not cause obvious issues."
[11:55] Ubit Umarov: yes ppl should replace those things
[11:55] Ubit Umarov: that option may become ON
[11:55] LaNani Sundara: ppl do not know they are "bad" things
[11:55] LaNani Sundara: when the thing renders
[11:55] Ubit Umarov: that is why it should be a option
[11:56] Ubit Umarov: so turn on/off allows test
[11:56] Bill.Blight perpetuating the idea that , just because it worked before , even though it was broken that it should still work, is a boat anchor around the neck of Opensimulator
[11:56] Ubit Umarov: but they may change the render engine, and then some may became deadly
[11:57] LaNani Sundara: hmm it is baggage, things uploaded in the past and still around
[11:57] Ubit Umarov: but if you look to old emails on mlists you may find ppl kicking dahlia
[11:57] Ubit Umarov: bc mesmerize does refuse some hollow sizes etc
[11:57] Ubit Umarov: ( and so does ubOdeMesh... )
[11:58] Ubit Umarov: a lot of viewers added "special prims" many bad
[11:58] LaNani Sundara: viewers added them?
[11:58] Ubit Umarov: yes
[11:58] Ubit Umarov: on the objects selector
[11:59] LaNani Sundara: ah so those are not standard things
[11:59] Ubit Umarov: if you notice now fs changes the max hollow acording to tis shape etc
[12:00] Ubit Umarov: in past some did not..
[12:00] Ubit Umarov: also..
[12:00] Ubit Umarov: and a hollow 2 large may cause the inside faces to be actually on the outside
[12:00] Ubit Umarov: for example
[12:00] LaNani Sundara: ah i see
[12:01] Ubit Umarov: that kind of bad lanani
[12:01] LaNani Sundara: ok thank you for explaining
[12:01] Ubit Umarov: then viewers do ipload broken meshes
[12:01] Andrew Hellershanks: The build dialog box in FS has some additional choices for starting prims. Most of those are just simple mods of standard prim types.
[12:01] Ubit Umarov: so guess that does some checks.. not sure
[12:01] Gavin.Hird my favortie was always the megaprims that if resized snapped back to a 10x10. That ruined a lot of builds
[12:02] Ubit Umarov: :)
[12:02] Andrew Hellershanks: Ah, yes. I remember those. :)
[12:02] Ubit Umarov: they where a hack to bypass limits
[12:02] Ubit Umarov: resize made the limits kick in
[12:02] Gavin.Hird absolutely
[12:02] LaNani Sundara: were they mesh?
[12:03] Andrew Hellershanks: There were at least two different ways to search for the mega prims so you could try and find the one you needed. Not all size combinations were available.
[12:03] Andrew Hellershanks: LaNani, no. This was long before mesh.
[12:03] Ubit Umarov: no lanani
[12:03] LaNani Sundara: ok
[12:03] Gavin.Hird tweaked prims
[12:03] Ubit Umarov: even before sculpts if i remember
[12:03] Andrew Hellershanks: It might have been.
[12:03] LaNani Sundara: so they were mega as in oversized?
[12:04] Ubit Umarov: just large prims lanani
[12:04] Andrew Hellershanks: LaNani, they were prims with at least one dimension that was greater than 10m.
[12:04] Ubit Umarov: 1000m etc
[12:04] OtakuMegane Desu: SL way back had a 10m limit on prim size. Megaprims were hacked in to get beyond that limit.
[12:04] LaNani Sundara: ah i arrigato
[12:04] OtakuMegane Desu: They ranged everywhere from 11m to things like 65535m
[12:05] Andrew Hellershanks: Many viewers also enforced the 10m limit in the build dialog box.
[12:05] Ubit Umarov: some one found a hack to bypass the limits checks and made packs of those until ll fixed the code
[12:05] Ubit Umarov: ll considered to remove them.. but gave up
[12:05] Gavin.Hird filling an entire sim with a 256 x 256 x 10 was a hobby for some for a while
[12:05] Andrew Hellershanks: The megaprims were passed around and used by a lot of people.
[12:06] Ubit Umarov: they do messup LOD
[12:06] LaNani Sundara: but now... you can just make a huge prim without a hack
[12:06] Ubit Umarov: well i still have a few packs of those :)
[12:06] Andrew Hellershanks: You used to see people in build related group asking if people had a megaprim with a given set of dimensions.
[12:06] Ubit Umarov: well not on some grids lanani
[12:07] Gavin.Hird there was a black market for megaprims hehe
[12:07] Andrew Hellershanks: I think I still have some megaprims in my SL inventory along with the tool(s) to search for other sizes.
[12:07] LaNani Sundara: funny
[12:07] Ubit Umarov: but most use the default limits
[12:07] Gavin.Hird I have them too
[12:07] OtakuMegane Desu: I have several sets
[12:07] George Equus: Me too, used some as sky platforms
[12:07] Ubit Umarov: :)
[12:07] OtakuMegane Desu: The old originals and from the second round when a glitch allowed it a second time.
[12:08] Andrew Hellershanks: We are just past the top of the hour. Any last minute topic for today before we wrap things up?
[12:08] Gavin.Hird I even think I tried to transfer them to OpenSim with SI in 2008, but they all ended up as 10x10
[12:08] Ubit Umarov: think i do have them here
[12:08] LaNani Sundara: reminds me how on Kaneva we learned how to break through the borders, they were closed barriers, and build outside
[12:08] Andrew Hellershanks: Gavin, it is possible that SI may have enforced the 10m limit on upload or that may have been after the LL code was fixed to enforce the limit.
[12:08] Ubit Umarov: but .. no need at all now on our regions :)
[12:09] Gavin.Hird I think it was SI that was the issue
[12:09] OtakuMegane Desu: Yeah. The old megaprims don't really have a use in OS.
[12:09] Gavin.Hird I sill have a copy on an old VM
[12:09] OtakuMegane Desu: Even in SL their uses dropped massively once SL let up some on the dimension limits.
[12:09] Andrew Hellershanks: Gavin, I still have a copy on the Windows side of my machine. I don't think SI works any more.
[12:10] Gavin.Hird I tried it a couple years back and it was able to log in
[12:10] OtakuMegane Desu: SI doesn't to my knowledge. I've tried a couple times in recent years, it won't even connect properly last time I tried.
[12:10] Ubit Umarov: my inv here has SI items
[12:10] Andrew Hellershanks: ok. Been quite a few years since I last tried to use it.
[12:11] Gavin.Hird something like the old Imprudence
[12:12] LaNani Sundara: still funny how Adrew's chairs have such low LOD
[12:12] Andrew Hellershanks: That will do it for this week. See you again next week.
[12:12] Andrew Hellershanks: LaNani, and how they often take so long to rez.
[12:13] Ubit Umarov: FS added something to reset avatar meshes LOD

Personal tools
About This Wiki