Chat log from the meeting on 2022-07-05

From OpenSimulator

Jump to: navigation, search
[11:01] Selby.Evans hi everyone
[11:02] Andrew Hellershanks: Hello, everyone.
[11:02] Ubit Umarov: hi
[11:03] Kayaker Magic: Hello all!
[11:03] Motoko.Karu hihi
[11:04] Xenon.Darrow hi!
[11:04] Andrew Hellershanks: Hello, Xenon
[11:04] Xenon.Darrow Is this the dev meet?
[11:04] Andrew Hellershanks: yes, it is.
[11:04] Andrew Hellershanks: Have a seat.
[11:04] Xenon.Darrow are we voicing or typing?
[11:04] Andrew Hellershanks: Typing
[11:04] Vincent.Sylvester Let's see, this week we had a fix for OARs not saving looped sounds or sounds in general I suppose. Some more cleaning on mantis. Think that's about it
[11:04] Selby.Evans Hi Xenon
[11:04] Xenon.Darrow Hey Selby!
[11:05] Andrew Hellershanks: Vincent, right. The second commit with the file that got missed was the important one that actually fixed the issue. :)
[11:05] Ubit Umarov: details :p
[11:05] Vincent.Sylvester You know it's an important fix when Ubit forgets a file
[11:06] Ubit Umarov: :)
[11:06] Andrew Hellershanks: hehe
[11:06] Ubit Umarov: and usually is the most relevant file, lol
[11:07] Kayaker Magic: I'm always afraid I'll grab a copy of the sources just before Ubit submits one with the comment "oops"
[11:07] Ubit Umarov: so what news do you hae abotu opensim ?
[11:08] Ubit Umarov: have... about...
[11:08] Andrew Hellershanks: Kayaker, quite possible it could happen. :)
[11:08] Andrew Hellershanks: Nothing much in the way of news this week other than what Vincent already mentioned.
[11:08] Ubit Umarov: Ooops commits are usualy important ones :)
[11:08] Vincent.Sylvester Did prompt me to look on mantis if there were other sound issues, actually found and closed one out. Still like two dozen more I need to test and update
[11:09] Andrew Hellershanks: The only other changes were minor code clean ups changing some "!...IsZero()" to "...IsNotZero()".
[11:09] Ubit Umarov: yeah minor thing.. just reads a little better
[11:10] Ubit Umarov: when jit is clever, code is basicly the same
[11:10] Andrew Hellershanks nods
[11:10] Ubit Umarov: a bit faster when it is not that clever :)
[11:12] Andrew Hellershanks: I was supposed to push some changes to the OpenSimProfile module last week but got side tracked with some other tasks. I have made a note as to which version of code requires the OSP changes. I hope to get those changes committed and pushed today.
[11:14] Vincent.Sylvester Similarly been meaning to finish up the work on search to give you that code to update that repo as well, but find myself knee deep in javascript hell the past two weeks
[11:14] Andrew Hellershanks: I try and avoid JS as much as possible. :)  Sometimes it is necessary.
[11:15] Vincent.Sylvester I thought python was bad, js takes the cake and throws it across the great lakes
[11:15] Andrew Hellershanks: I also will check whether the changes to the profiles affects the search module. I don't think it does as I don't recall having any compile issues with OpenSimSearch.
[11:17] Vincent.Sylvester Later this week I'll give mono a good looking at rolling back to older release given the new one seems to be troublesome
[11:18] Vincent.Sylvester Haven't been able to go through all the commits between the two, either too much code changed or too complicated to read without knowing the codebase
[11:18] Andrew Hellershanks: Vincent, is that the 6.12 giving trouble?
[11:18] Vincent.Sylvester Pretty much the day after the 179 release when I upgraded problems kept rolling in
[11:19] Vincent.Sylvester Always same thing, I did mantis what I have but no conclusive data
[11:19] Ubit Umarov: vicent, andrew is still on the big merge commits review
[11:19] Ubit Umarov: ;)
[11:19] Vincent.Sylvester I got to 143 then my brain stopped working
[11:19] Ubit Umarov: andrew has like 2000 still ?
[11:19] Andrew Hellershanks: Ubit, hehe... yea, one of these days I may finish going through that list. It is around 2k still to go through.
[11:20] Ubit Umarov: :)
[11:20] Vincent.Sylvester If it turns out rolling back resolves the issues I'ma be a bit cross with mono
[11:20] Andrew Hellershanks: After that will be reviewing all the other changes made since the big merge.
[11:20] Ubit Umarov: well Seems OpensimFest opens next friday
[11:20] Ubit Umarov: guess those are other 2k set :)
[11:22] Andrew Hellershanks: Vincent, I have the 6.12 182 mono installed and haven't really noticed any issues. Perhaps because I haven't been running my Standalone much in recent days.
[11:22] Andrew Hellershanks: Ubit, possibly.
[11:23] Vincent.Sylvester The more stuff on the region the worse it gets, you can read up on the mantis what it looks like, very strange thing and technically I suppose an impossible crash in a way
[11:23] Vincent.Sylvester Randomly happens if people are on it otherwise its fine, does a datasnapshot then stacktraces on a uuid tryparse in some lsl list function
[11:24] Vincent.Sylvester Really strange thing
[11:24] Andrew Hellershanks: Vincent, remind me which mantis that is as I have some testing to do this week for compatability with some scripts in X and Y engines.
[11:24] Vincent.Sylvester 9012
[11:24] Xenon.Darrow Apologies, I have a work emergency that popped up. :D
[11:24] Andrew Hellershanks: That is an odd place for an issue.
[11:24] Andrew Hellershanks: Xenon, np.
[11:25] Vincent.Sylvester It's crashing in a try catch, now I don't know the ins and outs of C#, but I thought that was so that wouldn't do that
[11:25] Andrew Hellershanks: Um... if it is inside a try catch it shouldn't crash.
[11:26] Ubit Umarov: ofc we also need to consider that Vicent machines did develop some kind of vodooo  :)
[11:26] Vincent.Sylvester It has started happening on more machines now so annoyingly not just bad hardware
[11:27] Ubit Umarov: Mono looks more like a Vodoo thing, that actual programming ;)
[11:27] Andrew Hellershanks: The stack trace in the mantis doesn't include line numbers. If it isn't easy to reliably trigger the bug it will be hard to track down the cause.
[11:27] Ubit Umarov: worse, does not tell reason for the trace
[11:28] Vincent.Sylvester I compiled in full debug, even libomv not sure why it doesn't provide line numbers
[11:28] Andrew Hellershanks: Right. That is an extract of the messages output by the OS instance.
[11:28] Vincent.Sylvester There is no other log, it does a datasnapshot then immediately crashes
[11:28] Vincent.Sylvester I cannot fathom how that triggers something inside script engine either
[11:29] Andrew Hellershanks: That raises the question as to whether it happens during the latter part of DS or if DS completed.
[11:29] Vincent.Sylvester I thought maybe datasnapshot puts a lock on something that then interferes with scripts, but I can't see anything that would do that
[11:30] Vincent.Sylvester Just doesn't fit together at the moment leaving me only the option that seems most obvious, updated mono problems start, so remove that variable first
[11:30] Ubit Umarov: well try again mono ....122
[11:31] Ubit Umarov: to see it is is mono
[11:31] Ubit Umarov: if...
[11:31] Ubit Umarov: you did compile libomv on 179 ?
[11:31] Vincent.Sylvester Yes
[11:32] Ubit Umarov: guess no dif, since that is a runtime issue
[11:32] Ubit Umarov: well mono changes from 122 to 179 seem meaning less
[11:33] Ubit Umarov: one full year of minor commits
[11:33] Ubit Umarov: ofc now they are like that opensim guy and have commits like "several bug fixes"
[11:33] Ubit Umarov: :p
[11:33] Vincent.Sylvester I haven't found anything relating to either try catch handling, crash avoidance or the operations done inside tryparse either
[11:34] Vincent.Sylvester It's one of these illusive things I seem to attract for some reason
[11:34] Andrew Hellershanks: Two backtraces reference YEngine. Does the problem also affect XEngine?
[11:35] Vincent.Sylvester I think I only have one region left with that, but I could clone the worst effected region onto that to see if that does anything
[11:35] Vincent.Sylvester Though they both in the end use the same tryparse
[11:36] Andrew Hellershanks: Might be worth doing to eliminate the script engine as being part of the problem.
[11:36] Vincent.Sylvester I wish I could turn that hash or whatever into a line number somehow to see which specific line it is it crashes on
[11:36] Vincent.Sylvester Even thought about just going into the lsl api and adding a try catch around all try parse
[11:37] Vincent.Sylvester Not sure that would do anything
[11:37] Andrew Hellershanks: Those are memory offsets in reference to code blocks.
[11:37] Vincent.Sylvester If I break open libomv dlls can I then use that to find the line?
[11:38] Andrew Hellershanks: hm... if the error is in libOMV it would need to be compiled with debugging information to get line numbers.
[11:39] Vincent.Sylvester I did that
[11:40] Andrew Hellershanks: Based on the information in the mantis I have no idea why it is not showing line numbers.
[11:40] Andrew Hellershanks: Vincent, Keep plugging away at it and let us know next week how you got on and what you manage to find out.
[11:41] Vincent.Sylvester If rollback helps I'll spin up some virtual machine to poke more at this on the current mono, add some more try catch everywhere, see if I get a different result
[11:41] Andrew Hellershanks: ok, sounds good.
[11:42] Vincent.Sylvester And then I'll tear mono a new behind
[11:42] Andrew Hellershanks: :)
[11:43] Andrew Hellershanks: We are getting close to the top of the hour. Jagga, you had a couple of questions.
[11:43] Jagga Meredith: Any idea why osTeleportAgent would be the only OSSL thingie broken on a region?  Fails silently unless you're the owner of the object.
[11:43] Ubit Umarov: only OSSL thingie broken on a region ?
[11:44] Jagga Meredith: all other OSSL code works
[11:44] Andrew Hellershanks: The wiki states -> For osTeleportAgent() to work, the owner of the prim containing the script must be the same as the parcel that the avatar is currently on.
[11:44] Jagga Meredith: ahhh
[11:44] Jagga Meredith: also, i lied
[11:44] Jagga Meredith: OSSL Permission Error: osGetSimulatorMemoryKB permission denied.(script: OSSL tester event: touch_start primID:c0d69483-8d32-42da-826b-0bba6530e0d6 at <4.347606, 251.5064, 21.25>)
[11:45] Jagga Meredith: this broke
[11:45] Ubit Umarov: rtfm ?
[11:45] Ubit Umarov: :p
[11:45] Andrew Hellershanks: The threat level of that is shown as Severe so permissions may also come in to play.
[11:45] Vincent.Sylvester You need to update your ini files on that then
[11:45] Jagga Meredith: ok
[11:45] Vincent.Sylvester Check osslDefaultEnable.ini
[11:46] Ubit Umarov: namely osslEnable.ini
[11:46] Jagga Meredith: ok
[11:46] Andrew Hellershanks: yes, the one that Ubit mentioned.
[11:46] Kayaker Magic: Doesn't ossEnable over-ride threat levels?
[11:46] Ubit Umarov: ofc
[11:46] Jagga Meredith: thaat's all i've got
[11:47] Andrew Hellershanks: ok.I thought you said you had two questions.
[11:47] Ubit Umarov: replaces levels by condtionls, like region owner, etc etc
[11:47] Jagga Meredith: that was2
[11:47] Vincent.Sylvester By default that's set to Allow_osTeleportAgent =           ${OSSL|osslParcelO}ESTATE_MANAGER,ESTATE_OWNER so the parcel owner, estate owner and estate managers can run that function in objects they own
[11:47] Andrew Hellershanks: Jagga, right. Hope that sorts out your issues.
[11:48] Jagga Meredith: perfect
[11:48] Ubit Umarov: threat level is a crude fallback when nothing else is defined on osslEnable
[11:48] Ubit Umarov: well fallback for ossl that do have a threat level..
[11:49] Andrew Hellershanks nods
[11:49] Ubit Umarov: as you know i did add several ossl that jsut have no threat at all
[11:49] Ubit Umarov: ie are just like LSL
[11:50] Andrew Hellershanks: Yes, not all OSSL functions have threat levels.
[11:50] Vincent.Sylvester There are a few silent failure points in there for when the agent isn't on the parcel owned by the object owner that initiates the teleport, if I read that correctly. Also bad region names, presence and other things will cause it to fail without saying anything
[11:50] Vincent.Sylvester Does the script work on a different region?
[11:50] Ubit Umarov: well this was a opensim policy change..  all ossl used to have threat levels :)
[11:51] Ubit Umarov: also from the ones that did think opensim should be a pure SL clone, even disabling all OSSL
[11:52] Ubit Umarov: well legacy things..
[11:52] Andrew Hellershanks: Right. The default threat level always allowed some of the OSSL functions that were at or below that level.
[11:52] Vincent.Sylvester In the end it's just names really
[11:53] Andrew Hellershanks: It is also possible to allow a given avatar to use a given OSSL function without making the function available for all to use.
[11:53] Vincent.Sylvester You can set avatar uuids, groups as well I think
[11:53] Andrew Hellershanks nods
[11:54] Andrew Hellershanks: I haven't tried it with groups.
[11:54] Andrew Hellershanks: About 5 minutes left in the hour. Any other questions for today?
[11:54] Vincent.Sylvester Just don't set them to true unless you want EVERYONE to have access to them
[11:55] Andrew Hellershanks: Vincent, exactly.
[11:55] Andrew Hellershanks: Kayaker, no questions from you today?
[11:56] Kayaker Magic: hehe, I'm good this time
[11:56] Andrew Hellershanks: ok. You have been rather question free the last few meetings. :)
[11:56] Jamie.Jordan :-)
[11:56] Kayaker Magic: busy in RL
[11:57] Andrew Hellershanks: oh, one other comment about the OpenSimProfile changes. I will do an initial commit to allow the module to compile. I still need to do additional testing regarding the handling of the publish flags.
[11:58] Andrew Hellershanks: Even if those flags aren't working 100% it isn't that big a deal as I suspect few if any grids are making use of them.
[11:59] Andrew Hellershanks: Any final questions/comments for today as we are now less than 1 minute from the end of the hour?
[11:59] Kayaker Magic: Quitting on time? That never happens!
[12:00] Andrew Hellershanks: For a moment I thought we might finish a minute early. :)
[12:00] Selby.Evans Bye all
[12:00] Andrew Hellershanks: ok, Selby.
[12:00] Andrew Hellershanks: Time to wrap up todays meeting. Thank you all for coming. See you again next week.
Personal tools
About This Wiki