Chat log from the meeting on 2017-11-14

From OpenSimulator

Jump to: navigation, search

[11:13] Arielle.Popstar how is kitely coming along with upgrading to .9?
[11:13] Arielle.Popstar are they intending to?
[11:14] Kayaker Magic: Kitely does not trust 0.9. Too many changes at once.
[11:14] Kayaker Magic: They are waiting for a lot more testing to happen.
[11:14] Arielle.Popstar oh
[11:14] Andrew.Hellershanks If everyone waits for more testing to happen no one will upgrade so no testing will be getting done.
[11:14] Arielle.Popstar osgrid been testing since it changed over
[11:15] Leighton.Marjoram It must be a lot of work for them as well with the high volume of changes made to 0.9
[11:15] Andrew.Hellershanks I still haven't had the time to finish work on the release notes for 0.9
[11:15] Unknown.UserUPUUI i have been seeing quite a bit of 0.9.1.x on OS grid
[11:16] Kayaker Magic: Yeah, it is getting tested. Kitely will come around one day.
[11:16] Arielle.Popstar wouldn't want to go back to .8.2 anymore
[11:16] Andrew.Hellershanks Last week we were talking about the mute list feature. As of now, Melanie has pushed the code changes from Avination to support mutelist.
[11:18] Andrew.Hellershanks I was about half-way through the code to support mutelist that used an external PHP file as part of the process but I have stopped work on that module. I haven't had a chance to test the Avination mutelist changes. I only discovered the code had been pushed about a few hours ago.
[11:19] Arielle.Popstar so Mel was holding the mutellist code back?
[11:19] Arielle.Popstar the avination merge didn't include all of it?
[11:20] Andrew.Hellershanks No, the Avination code merge did not include the mutelist code changes.
[11:20] Sheera Khan: Git says, it was a seperate repository so it didn't get merged
[11:20] Leighton.Marjoram
[11:21] Arielle.Popstar oh so not complete
[11:21] Sheera Khan: I really appreciate the speed with which this feature was worked upon ;-)
[11:21] Andrew.Hellershanks We now have two mutelist related modules in the source tree. The old one that didn't really do anything might get removed after a while. That is something I will need to talk about with the other devs. Doesn't seem much point in keeping the old mutelist module.
[11:21] Andrew.Hellershanks Sheera, sometimes you get lucky. Sometimes you don't. :)
[11:22] Sheera Khan: so sorry to hear that your work stopped bc an already existing Blob was committed
[11:22] Andrew.Hellershanks Sheera, np. My work would have been done as an external module. The core code will not include features that require support from external PHP files.
[11:23] Sheera Khan: just so I get that right:
[11:24] Sheera Khan: the current code isn't done yet there is still some work needed to fully port it to OpenSim
[11:24] Sheera Khan: when it is done it requires changes in the Robust server
[11:25] Sheera Khan: I didn't get the right word fast enough for writing in open chat
[11:25] Arielle.Popstar i am just curious as to what other changes are not merged yet
[11:25] Andrew.Hellershanks Sorry, I don't think I'm following you. What is it you think isn't done yet that may require changes in Robust?
[11:27] Sheera Khan: just a moment, I'm looking it up...
[11:27] Leighton.Marjoram I was wondering if there is a feature matrix that shows what is supported by the core mute list and what features will require a php external file?
[11:27] Arielle.Popstar this part maybe Sheera? maybe someone can step up to make it work in an open grid context.
[11:27] Sheera Khan:
[11:27] Sheera Khan: next to last note reads:
[11:28] Sheera Khan: I've dropped the old Avination mute code. It's database needs to be adapted to use Robust services because it requires a direct MySQL connection as it is now.
[11:29] Andrew.Hellershanks Leighton, there would have been no difference. There are only three main messages that a mutelist module needs to handle. I already have my module adding/removing mutelist requests in a database. The version in core now does what I was close to finishing without the need for any external PHP files.
[11:29] Leighton.Marjoram oh ok that makes sense to me thank you Andrew
[11:29] Andrew.Hellershanks yw, Leighton.
[11:30] Andrew.Hellershanks Sheera, I read that too. I'm not sure what that really means. I haven't had time to study the code changes.
[11:30] Andrew.Hellershanks I do know that a file was added that I think may be handling some of the database operations needed to support the mutelist module.
[11:31] Sheera Khan: It sounded like there are changes to the database and robust so the whole grid needs to upgrade
[11:31] Andrew.Hellershanks There are changes to the database as a new table is needed to store the mutelist information.
[11:31] Sheera Khan: most of the stuff seems to be handled by the region though
[11:32] Andrew.Hellershanks Sheera, right. That is why I'm not quite sure where Robust enters the picture re: mute information.
[11:32] Arielle.Popstar would need it to be with robust if available on login
[11:33] Arielle.Popstar grid wide\
[11:33] Ubit Umarov: hm dont ask me about new mutes code... i just arrived home and had not seen it :)
[11:34] Andrew.Hellershanks The mutelist may be getting sent when a person logs in. I'm not sure if that is handled by Robust or when you first enter a region. After that, the viewer has the information cached.
[11:35] Andrew.Hellershanks ubit, I haven't looked at it in any detail. It only hit master about two and a quarter hours ago, afaict.
[11:35] Sheera Khan: the log message tells me the viewer requests the mute list from the region
[11:35] Ubit Umarov: i didn't pulled.. arrived and rushed here :)
[11:36] Ubit Umarov: well as before that commit
[11:36] Ubit Umarov: viewer does ask for a list
[11:36] Ubit Umarov: region returns one ( empty so far) or a use cache message
[11:36] Andrew.Hellershanks There is a crc check when the mutelist is requested. That allows a decision to be made whether to send the full mutelist or to continue using what information the viewer has already cached.
[11:37] Sheera Khan: initializeRequest: Requesting xfer from for file:
[11:37] Ubit Umarov: that melanie comment on mantis is odd... can't be like that
[11:37] Sheera Khan: 9051 is the region port
[11:37] Andrew.Hellershanks yea, sounded a bit odd to me too but I haven't looked in to the DB side of the code.
[11:37] Ubit Umarov: ( and does not match some old coments on code that said offline would not work )
[11:38] Bill.Blight Not to mention just plain RUDE ...
[11:38] Ubit Umarov: sheera, that entry is with current code.. a empty file is sent to viewer.. that is it getting it
[11:38] Andrew.Hellershanks sorry, I was wrong. There is no additional file handling the DB side of maintaining the mutelist.
[11:39] Andrew.Hellershanks Not one that was added in the recent commit.
[11:39] Ubit Umarov: but now we need to look to the new code mel gave us
[11:39] Sheera Khan: ;-)
[11:39] Sheera Khan: at least there's always a bit to do ^^
[11:40] Andrew.Hellershanks Ubit, I was saying earlier that I had already partially gotten mutelist working using a PHP helper. I've stopped work on that for now. It has meant I've been looking at what needs to be done in code.
[11:40] Ubit Umarov: i know
[11:40] Ubit Umarov: but that now seems outdated
[11:42] Andrew.Hellershanks I will test the code later. I've been dealing with a bug in KiCad that last few days. I've finished that (for now) so I can get back to OpenSim.
[11:44] Andrew.Hellershanks One announcement for today.
[11:44] Andrew.Hellershanks We are now just a few weeks away from this years running of the Open Simulator Community Conference.
[11:45] Kayaker Magic: AHHH!
[11:45] Andrew Hellershanks: It is being held on December 9 and 10. The web site with details about this years event can be seen at
[11:46] Bill.Blight Pre-Party the night of the 8th on my grid, it's already on the OSCC calendar ... hehe
[11:46] Bill.Blight SingerGirl show
[11:46] Andrew Hellershanks: I will check whether someone involved with the conference is available who could attend the meeting and provide more details about this years event.
[11:46] Andrew Hellershanks: Sounds good, Bill.
[11:47] Selby.Evans Virtual Worlds Fair has an initial site: OSCC, Thirza's HG exhibit: : A project of the HG Chamber of Commerce
[11:47] Andrew Hellershanks: The location of the conference can be reached via direct login to the region or via HG TP.
[11:47] Selby.Evans HG address given there
[11:48] Selby.Evans (In the article)
[11:48] Andrew Hellershanks: ty, Selby. It will also be on the conference website.
[11:48] Selby.Evans HG address: Expo Zone 2
Put the HG address in the Find bar of your world map. Click Find, TP
Get a landmark when you are there.
[11:49] Selby.Evans that is the expo zone -- you can go other places from there
[11:50] Andrew Hellershanks: If you plan to attend the conference you should register for it. The conference is well attended. By registering you will get assigned entry to one of the four keynote areas.
[11:50] Andrew Hellershanks: The display areas are open to everyone.
[11:51] Selby.Evans The display areas will be open the year round
[11:52] Kayaker Magic: I have a booth there in Expo 4.
[11:52] Kayaker Magic: It is cheep advertizing for my Kitely store, I can leave stuff up there all year.
[11:54] Andrew Hellershanks: This hour has flown by. Any other topics for today?
[11:54] Kayaker Magic: On a subject from last week, I tried to do some timing tests of rezzing objects with different MySQL settings.
[11:55] Kayaker Magic: The short version of the story is: If you run a grid, you should have someone tune your database who knows what they are doing. (That person is NOT me).
[11:55] Andrew Hellershanks: Kayaker, you were finding that some DB settings have a big impact on rez times?
[11:55] Kayaker Magic: Yes,
[11:55] Kayaker Magic: MySQL has several internal algorithms you can choose from ISM is one one, and that works well for small databases and small grids. INNODB is another internal MySQL method that works better for larger databases.
[11:56] Kayaker Magic: If you switch from ISM to INNODB you may see a HUGE slowdown in rezzing some objects or uploading mesh. This is caused by the default value of a MySQL parameter. You can fix this by setting innodb_flush_log_at_trx_commit =0 in the mysqld.cnf file.
[11:56] Ubit Umarov: ok ppl i did look a bit to the new Xmutes code
[11:56] Bill.Blight I use MariaDB with the Aria engine, great speed improvement
[11:56] Ubit Umarov: and it is only good for standalones and closed grids
[11:56] Ubit Umarov: ( from what i see )
[11:57] Kayaker Magic: Bill: I think you can get that same speed improvement by tuning MySQL for OpenSim.
[11:57] Ubit Umarov: it depends on regions direct access to central mySql DB
[11:57] Andrew Hellershanks: Kayaker, the mutelist table would be a small table so it would make sense to use ISAM for it.
[11:58] Andrew Hellershanks: Kayaker, I think we talked about innodb_flush_log_at_trx_commit =0 last week (or was it a similar that was discussed?). Is that the one that has risks associated with it being set to 0?
[11:59] Ubit Umarov: ( that was fine at Avination, and will be ok on similar grids that only run own regions )
[11:59] Ubit Umarov: should also be ok for standalones
[11:59] Bill.Blight That is a maybe , kayaker, but I already got greater speed using Aira, and it is , IMHO, a much better engine.
[11:59] Kayaker Magic: No, there is little risk with that one. I was tinkering with a different parameter the week before that was very risky.
[12:00] Andrew Hellershanks: Kayaker, ok. I just wanted to clarify that.
[12:01] Andrew Hellershanks: Kayaker, What is the default value and how much of a change did you notice when you set it to 0?
[12:02] Kayaker Magic: Some mesh that failed to upload at all now uploads in seconds. Some compound objects that were slow to rez now rez 200 times faster.
[12:02] Kayaker Magic: But it is a live grid and nobody is interested in slowing it down to do tests.
[12:02] Andrew Hellershanks: ok. That's some big differences.
[12:02] Andrew Hellershanks: I have a couple of other places where I can test that.
[12:03] Kayaker Magic: And my server is on OSGrid and testing there doesnt help.
[12:03] Andrew Hellershanks: No, you wouldn't be able to change the DB setting.
[12:11] Ubit Umarov: about xmutes, mel did confirm what i said
[12:12] Ubit Umarov: Xmutes as they are require regions to direct connect to central DB
[12:12] Sheera.Khan oh, no grids for now ...
[12:12] Ubit Umarov: so... ok for standalones and closed grids ( where that is not a issue )

Personal tools
About This Wiki