Chat log from the meeting on 2019-06-25

From OpenSimulator

Jump to: navigation, search

[11:06] Andrew.Hellershanks @hg.osgrid.org: I can start us off today with a question.
[11:06] Kayaker Magic: Before Ubit arrives and gets mad at me for mentioning it, does anyone know when Nanii forked off? Was it before or after YEngine?
[11:07] Andrew.Hellershanks @hg.osgrid.org: Have any of you been using the DTL/NSL money module with OS 0.9? I was talking to a grid owner who has that combo and they are having a problem where NPCs and HG visitors lock up the money system until a local person enters the region.
[11:07] Bill.Blight @hg.osgrid.org: Originally it was before Yengine but after XMREngine, I think ... (before it really worked and was renamed)
[11:08] Sheera Khan: Metro doesn't have a money module active...
[11:09] Bill.Blight @hg.osgrid.org: DTL/NSL had to be updated for current dev branch, I made it work for testing but I have since forgotten what I had to fix .. but yes there were issues if just using it stock
[11:09] Sheera Khan: except for some Gloebits and private owned regions I don't have any experience with ...
[11:09] Kayaker Magic: Yes, on Discovery Grid they are using DTL and it has a problem that they fixed. There was talk of publishing the fix but there are so many things in the to do list....
[11:09] Andrew.Hellershanks @hg.osgrid.org: The combination used to work in 0.8 and I know there have been changes to the money system API and also changes related to NPCs. I suspect the module needs some sort of modification. I haven't looked in to what exactly.
[11:09] Gavin.Hird @grid.xmir.org:8002: unless the module has been changed to work properly with 0.9.x I guess there would be issues
[11:10] Andrew.Hellershanks @hg.osgrid.org: ok. From what Kayaker is saying other grids using that combination of OS and module have had a similar. or same problem.
[11:11] Kayaker Magic: The bug DiscoveryGrid had was this: DTL doesn't disable itself when there is another money module like Gloebits selected. Do they could not have some regions on DTS and some on Gloebit.
[11:11] Andrew.Hellershanks @hg.osgrid.org: The DTL/NSL module is old and hasn't been updated for some time.
[11:11] Andrew.Hellershanks @hg.osgrid.org: Ok. That is a different problem entirely.
[11:13] Andrew.Hellershanks @hg.osgrid.org: It is easy to add a setting to enable or disable a module.
[11:13] Sheera Khan: that mostly looks like a configuration error by the simulator owner
[11:14] Andrew.Hellershanks @hg.osgrid.org: Possibly. I have the source for that module but haven't tried setting it up on my machine. It isn't one of the simpler things to set up.
[11:15] Kayaker Magic: The problem was even if you changed the configuration settings, if the DTS module was there, it got loaded and did things when it was disabled in configuration.
[11:15] Ubit Umarov: hmm DTS needs special version for 0.91
[11:16] Kayaker Magic: Apparenly several different grids have independantly fixed bugs in DTS
[11:16] Ubit Umarov: ie updated version :)
[11:16] Andrew.Hellershanks @hg.osgrid.org: It needs some updates. I was just asking if anyone here has used it with 0.9 as I have a grid owner telling me they have run in to a problem. It is something they will need to get fixed as the module isn't part of core OS but I was hoping someone here had run across the problem.
[11:17] Andrew.Hellershanks @hg.osgrid.org: Kayaker, I wouldn't be surprised. The people who write it don't seem to have touched it in ages and I don't know if they have a way to submit bug fixes.
[11:17] Kayaker Magic: I recall that it just needed re-compiling to work with 0.9.1
[11:18] Ubit Umarov: it needs more than recompile for 0.9.1 i think
[11:18] Bill.Blight @hg.osgrid.org: I needed changes to the prebuild and some of the includes, and a recompile
[11:18] Bill.Blight @hg.osgrid.org: if I remember correctly
[11:18] Ubit Umarov: and some uses of sll
[11:18] Ubit Umarov: ssl
[11:18] Andrew.Hellershanks @hg.osgrid.org: I agree with Ubit. The person who reported the problem should have already tried that. I'll double check to make sure, just in case they haven't.
[11:18] Bill.Blight @hg.osgrid.org: yes
[11:18] Bill.Blight @hg.osgrid.org: that too
[11:18] Ubit Umarov: the cert validation
[11:18] Ubit Umarov: it cant over ride the global rules
[11:19] Ubit Umarov: now that is possible with .net4.6, well easier at least
[11:19] Andrew.Hellershanks @hg.osgrid.org: Ubit, the problem reported is that the money module stops working when an NPC is rezzed or when an HG visitor enters the region. The money module works again when a regular local avatar enters the region.
[11:19] Ubit Umarov: the validation can bow be by endpoint
[11:20] Andrew.Hellershanks @hg.osgrid.org: I suspect it gets hung up trying to access some data for an NPC or HG avatar that it can't get and gets stuck.
[11:20] Ubit Umarov: hmm mb npcs don't have something it wants.. don't see what
[11:20] Andrew.Hellershanks @hg.osgrid.org: I think I'm repeating myself in a sentence again. :)
[11:21] Ubit Umarov: and HG, well its HG :)
[11:21] Andrew.Hellershanks @hg.osgrid.org: I'm not sure either. It should be simple enough to modify the module to ignore balance requests for NPCs. Not sure how easy it is to detect an HG visitor.
[11:22] Ubit Umarov: we should had a simple set of flags for that
[11:22] Andrew.Hellershanks @hg.osgrid.org: That would be helpful.
[11:22] Andrew.Hellershanks @hg.osgrid.org: You can add that to your ToDo list, Ubit.  ;)
[11:22] Ubit Umarov: but.. HG made as kinda external option makes that hard
[11:22] Cuga Rajal: Some HG visitors use My Suitcase, some not, is that a factor?
[11:23] Andrew.Hellershanks @hg.osgrid.org: Cuga, I don't know. I wouldn't expect it to affect the money system but as Ubit said. It is HG. :)
[11:24] Ubit Umarov: not related to money, but the try to make HG as normal users was kinda excessive.. somethings will never work
[11:24] Andrew.Hellershanks @hg.osgrid.org nods
[11:24] Ubit Umarov: HG travel should be a special mode
[11:25] Ubit Umarov: wel suitcase is part of that, but more things
[11:25] Cuga Rajal: Sounds complicated
[11:26] Ubit Umarov: HG ppl should travel light, as RL air travel :)
[11:26] Gavin.Hird @grid.xmir.org:8002: only with tourist visa and not full citizenship
[11:26] Andrew.Hellershanks @hg.osgrid.org: :)
[11:26] Ubit Umarov: yeap
[11:27] Kayaker Magic: I've heard reports of people moving all their inventory into the Suitcase so they don't have to deal with inventory when traveling.
[11:27] Gavin.Hird @grid.xmir.org:8002: usually you have to pay for excess luggage
[11:28] Andrew.Hellershanks @hg.osgrid.org: Kayaker, yikes.
[11:29] Bill.Blight @hg.osgrid.org: It's even more of a pain when you get several from a grid that does not use the suitcase landing in rapid succession ..
[11:29] Ubit Umarov: no idea why they can have edit rights on target region for example
[11:29] Cuga Rajal: Like OSG <clears throat>
[11:30] Andrew.Hellershanks @hg.osgrid.org: Not using a suitcase is partly a brower issue?
[11:30] Ubit Umarov: suitcase is also a problem thing.. not light at all
[11:30] Ubit Umarov: since it is a hack, it needs to fully replace viewers idea of the current inventory
[11:31] Kayaker Magic: Kitely implemented a kludge where they only load the first 1000 items in the suitcase when you arrive.
[11:31] Ubit Umarov: and scan for needed assets.. blabla
[11:31] Andrew.Hellershanks @hg.osgrid.org: Sounds like a good idea if there are people putting their entire inventory in to the Suitcase folder.
[11:31] Cuga Rajal: Its more of a security feature than a lightening of load..
[11:31] Ubit Umarov: not a light thing currently
[11:32] Ubit Umarov: but its a security thing mainy
[11:32] Ubit Umarov: to only expose those assets
[11:32] Gavin.Hird @grid.xmir.org:8002: so if some of your body parts and outfit are from the remaining, you end up looking like a plucked chicken?
[11:32] Ubit Umarov: to other grids
[11:32] Ubit Umarov: bc normal inventory is not transfered
[11:32] Cuga Rajal: I thought worn items always travel...
[11:33] Ubit Umarov: with normal inventory the host region asks for assets etc to source grid
[11:33] Ubit Umarov: and that is the problem
[11:33] Ubit Umarov: for normal inv, grids need to expose assets servers to world
[11:34] Ubit Umarov: when it should be only a subset
[11:34] Ubit Umarov: exportable things etc
[11:34] Cuga Rajal: Could add a suitcase enforcer option - reject HG avis without suitcase?
[11:34] Bill.Blight @hg.osgrid.org: that will make the "everything should be free" people cry ..
[11:35] Ubit Umarov: well first, understand how it (tries) to work is a idea :p
[11:35] Kayaker Magic: But OSGrid does not have suitcase!
[11:35] Kayaker Magic: So OSGrid travelers would all be rejected.
[11:35] Cuga Rajal: Thats one of the reasoins I rarely log with my OSG avi
[11:35] Ubit Umarov: osg had issues with suitcase
[11:36] Ubit Umarov: don't know the details
[11:36] Cuga Rajal: I was wondering about that, why that didnt support
[11:36] Gavin.Hird @grid.xmir.org:8002: they need to do a DB conversion
[11:36] Kayaker Magic: I'm told Dan doesn't like suitcases, doesn't think they add security.
[11:36] Cuga Rajal: I actually can access my non-suitcase stuff when HG teleported
[11:37] Cuga Rajal: So maybe fixes still needed therer for suitcase?
[11:37] Ubit Umarov: is there anything on opensim not needing fixes? :)
[11:37] Cuga Rajal: :)
[11:37] Bill.Blight @hg.osgrid.org: right on the nose there Ubit
[11:38] Gavin.Hird @grid.xmir.org:8002: so much for being the test grid
[11:38] Cuga Rajal: Just question of relative importance I guess
[11:38] Bill.Blight @hg.osgrid.org: that is my main complaint against Osgrid these days, less of a test grid and more wanting to win the popularity contest ...
[11:38] Ubit Umarov: osg is also a full grid, that limits testing
[11:39] Ubit Umarov: for example gird side testing is a problem
[11:39] Ubit Umarov: not good having a test destroy the dbs :)
[11:40] Cuga Rajal: You cant roll back in production
[11:41] Bill.Blight @hg.osgrid.org: osgrid just allowed itself to get too far away from core code , making it harder to compare and test, as far as grid side is concerned
[11:41] Gavin.Hird @grid.xmir.org:8002: of course you can, but it means downtime
[11:41] Ubit Umarov: not easy with TB dbs that take ages to backup and restore
[11:41] Cuga Rajal: downtime and loss of work ryv
[11:41] Bill.Blight @hg.osgrid.org: Yeah crazy big db's are no fun to restore
[11:41] Cuga Rajal: They really need a test version with separate db instance
[11:41] Gavin.Hird @grid.xmir.org:8002: and the core code only
[11:42] Ubit Umarov: we should had a test grid only
[11:42] Bill.Blight @hg.osgrid.org: That's what Ubit uses me for ... :P
[11:42] Ubit Umarov: but then, not one whould be there :)
[11:42] Ubit Umarov: and some issues need real load
[11:42] Cuga Rajal: You could never simlate the load, but a few beta testers is better than... 1
[11:42] Ubit Umarov: but not all
[11:43] Ubit Umarov: regions testing is a bit easier
[11:43] Ubit Umarov: lbsa is usually close to last bugs
[11:43] Kayaker Magic: Actually, OSGrid without a suitcase has shown some design problems with HG and suitcases, that was a good test.
[11:43] Kayaker Magic: Now if we could find a good fix....
[11:43] Andrew Hellershanks: The OSCC regions have been used to do load testing in the past. Unfortunately they haven't been updated in a while.
[11:44] Ubit Umarov: those tests where only for oscc usage
[11:44] Bill Blight: It is not like people have not offered to help update OSCC .....
[11:44] Ubit Umarov: ie to test if regions whould hold the meetings
[11:44] Gavin.Hird @grid.xmir.org:8002: with justincc updating the code realtime
[11:45] Andrew Hellershanks: Mostly true. They certainly helped to test the ability to handle large number of avatars in a region.
[11:45] Ubit Umarov: they needed many avatars per region :)
[11:45] Andrew Hellershanks: They certainly do
[11:46] Cuga Rajal: Thats really important and hard to test otherwise
[11:46] Ubit Umarov: and guess that a point 0.8 wasn't able to suport that many
[11:46] Ubit Umarov: ( or 0.7 ?? )
[11:46] Andrew Hellershanks: Is nebadon the main person looking after the OSCC regions?
[11:46] Bill Blight: Marcus
[11:46] Ubit Umarov: so justin did made a effort to improve that
[11:46] Bill Blight: Marcus pretty much was the go to guy last year
[11:47] Ubit Umarov: well that is a running effort.. still far from great
[11:47] Andrew Hellershanks: Bill, ok thanks. Probably time to talk to Marcus about updating the grid before it gets too close to the next OSCC and no time to do any update and testing.
[11:47] Cuga Rajal: MySQL DB tuning is a factor too Im sure, sure would be nice to have some info on OS website
[11:48] Ubit Umarov: funny ppl telling 0.91 does not suport what 20 avatars?
[11:48] Cuga Rajal: DB info on website is so old...
[11:48] Bill Blight: I"ve offered 3 years in a row, not going to offer to help again ... LOL :P
[11:48] Andrew Hellershanks: Cuga, you are welcome to add information. :)
[11:48] Ubit Umarov: and we do see a bit more at event plaza and similar regions :)
[11:48] Cuga Rajal: I', up for it
[11:48] Cuga Rajal: But I only have a small region with no visitors LOL
[11:49] Cuga Rajal: So hard to test tuning
[11:49] Gavin.Hird @grid.xmir.org:8002: host a crash party
[11:49] Cuga Rajal: Best written by someon on a grid that gets pounded
[11:49] Andrew Hellershanks: Cuga, if you are running a region attached to osgrid you wouldn't have access to the MySQL database to do any tuning.
[11:49] Cuga Rajal: My region is standaline HG not OSG
[11:50] Andrew Hellershanks: Cuga, ah, ok
[11:50] Cuga Rajal: Im a server engineer in RL
[11:50] Bill Blight: we had 51 at our fest test party, and it did not crash
[11:50] Bill Blight: and everybody was dancing away
[11:50] Ubit Umarov: .51 running on a Z80 ? :p
[11:50] Cuga Rajal: Thats the time to run DB testss, during partys :)
[11:51] Andrew Hellershanks: Just noticed the time. I may have to go near the top of the hour. Does any one have a question, comment, or topic of discussion for today that we haven't covered?
[11:51] Andrew Hellershanks: Ubit, Z-80? :)
[11:51] Bill Blight: haha
[11:51] Ubit Umarov: well if there was a 0.51 it was already some kinda of pentium :)
[11:51] Bill Blight: Ti-99 4a
[11:52] Andrew Hellershanks: Bill, I have a TMS 9900 CPU in my parts drawer. That can lead to a discussion for another time.
[11:52] Ubit Umarov: there are a few opened issues on mantis, i hope to look to them soon
[11:52] Cuga Rajal: I'll have a party this summer and do some DB tests :) Then I can help update the DB page
[11:52] Gavin.Hird @grid.xmir.org:8002: PDP-8
[11:52] Ubit Umarov: a bit out, as you may had noticed
[11:52] Ubit Umarov: that yengine control of memory usage is still nasty
[11:52] Bill Blight: I have a 8088 expansion daughter board for an old NEC I think in my closet ..
[11:53] Kayaker Magic: I'd like to ask Bill and Ubit about the advisability of using YEngine in a production environment.
[11:53] Bill Blight: my whole grid runs YEngine
[11:53] Andrew Hellershanks: Kayaker, I think you just did. :)
[11:53] Kayaker Magic: I've tested a few large scripts on YEngine with few problems.
[11:53] Bill Blight: so did the fest grid
[11:53] Bill Blight: there are a few gotchas
[11:53] Bill Blight: but simple work arounds
[11:54] Bill Blight: Till Ubit fixes it
[11:54] Bill Blight: LOL
[11:54] Ubit Umarov: the main issue i see now is that memory usage accounting
[11:54] Kayaker Magic: Any hope of saving the script state crossing from X to Yengine regions?
[11:54] Ubit Umarov: may stop some scripts for no valid reason
[11:55] Ubit Umarov: X to Y is working +-
[11:55] Ubit Umarov: not the reverse
[11:55] Bill Blight: That is the only issue I run into , scripts not cleaning varibles/memory, but an extra line to clear the varible is a simple workaround until it gets fixed
[11:55] Kayaker Magic: Adoption of YEngine would be easier if you could do it one region at a time. Can't do that if scripts reset on a crossing.
[11:55] Ubit Umarov: well y - > is not a easy thing
[11:55] Andrew Hellershanks: Ubit, does, or will, YEngine report actual memory used when using the LSL function to request memory use?
[11:56] Ubit Umarov: thats still bad andrew
[11:56] Cuga Rajal: brb
[11:56] Andrew Hellershanks: Ubit, ok.
[11:56] Ubit Umarov: a open mantis on that
[11:56] Bill Blight: It may not be accurate but it gets you in the ballpark
[11:56] Ubit Umarov: it "leaks" memory
[11:56] Andrew Hellershanks: Kayaker, Going from X to Y is moving to the future. We all know you can't go back in time. ;)
[11:57] Ubit Umarov: sometimes i just think removing it :p
[11:57] Ubit Umarov: doing that account does cost a lot
[11:57] Bill Blight: yeah that mantis is the engine not clearing variables when they are replaced
[11:57] Andrew Hellershanks: Makes a change from having memory always reported as 16384 regardless of actual use
[11:57] Ubit Umarov: its not variables or true memory usage, its just the account of it
[11:58] Ubit Umarov: well and strings passed as arguments
[11:59] Ubit Umarov: but that mem account is only usefull on a SL mainland model
[11:59] Ubit Umarov: or regions with rented parcels
[11:59] Ubit Umarov: not that usefull on own sims
[11:59] Ubit Umarov: not sure it pays the cost of having it..
[12:00] Ubit Umarov: btw ignore all Yengine extra options.. those are still not defined
[12:01] Ubit Umarov: swith, case, break, are sure to stay
[12:01] Kayaker Magic: I'll set up an X and YEnging region next to each other and try sailing back and forth between them.
[12:01] Kayaker Magic: In OSGrid.
[12:01] Ubit Umarov: from y -> X boat will stop
[12:02] Andrew Hellershanks: Kayaker, make sure you are wearing your life vest for when the boat stops suddenly and you get tossed in to the water. ;)
[12:03] Kayaker Magic: LOL
[12:03] Ubit Umarov: and not easy to fix that
[12:03] Kayaker Magic: This is a new mesh PFD I am wearing now!
[12:03] Ubit Umarov: Y state files are more compact and do not have all info needed to recover for X
[12:04] Ubit Umarov: for example variables names
[12:04] Andrew Hellershanks: Ubit, you just answered a question I was about to ask.
[12:04] Ubit Umarov: Y knows the order they are stored on file, no need for names
[12:04] Ubit Umarov: X is a xml like thing
[12:05] Ubit Umarov: well its xml :)
[12:05] Cuga.Rajal @rajal.org:9000: If you HG TP from Y to X region would scripts stop on worn items?
[12:05] Cuga.Rajal @rajal.org:9000: Or is it just physics iterms?
[12:05] Ubit Umarov: you lose state
[12:06] Cuga.Rajal @rajal.org:9000: Ah ok, so not just region crossing issue
[12:06] Ubit Umarov: for example global vars go back to compile values, etc
[12:07] Bill Blight: if you go from Y to X you may end up with the script returning to it's old state that it had on that region previously
[12:07] Cuga.Rajal @rajal.org:9000: effectively llReset
[12:07] Cuga.Rajal @rajal.org:9000: So only an issue if theres as state to pass that diff from original
[12:08] Cuga.Rajal @rajal.org:9000: If Y->X and it's a different X region that you started at, wouldnt restore?
[12:09] Bill Blight: I started storing the script settings in a notecard or in the description field, so even on lost state or reset they go back to how they were ...
[12:09] Kayaker Magic: Question: Are region crossings for attachment scripts more reliable in YEngine? XEngine has a race condition that messes up HUD scripts, for example.
[12:09] Cuga.Rajal @rajal.org:9000: You would have to write state changes to the NC
[12:10] Bill Blight: have had no isses with crossing Y TO Y
[12:10] Bill Blight: there is another Y region next door if you want to test
[12:10] Ubit Umarov: and Y does the transfer inside a event processing ( events have checkpoints for that)
[12:11] Ubit Umarov: while X just kills current event processing
[12:12] Ubit Umarov: ie on Y to Y execution is resumed more closely to the breakpoint
[12:13] Ubit Umarov: well not a major thing if you keep each event small :)
[12:14] Ubit Umarov: as you must on X, and should on Y
[12:15] Andrew Hellershanks: Event handlers should be small. Didn't stop one person from having to increase time allowed in an event handler because they ran in to a timeout issue otherwise. They were (are still) doing waaay too much in the handler.
[12:16] Andrew Hellershanks: Its quarter past. Probably time to wrap this up.
[12:16] Ubit Umarov: also only an event runs at a time per script

Personal tools
General
About This Wiki