[11:03 AM] Andrew Hellershanks: Hello, everyone.
[11:03 AM] Lyr.Lobo @cc.opensimulator.org:8002: Hello, Andrew!
[11:03 AM] Lyr.Lobo @cc.opensimulator.org:8002: our last week to meet before Europe moves forward an hour
[11:04 AM] Ubit Umarov: see eu is great.. can move time
[11:04 AM] Lyr.Lobo @cc.opensimulator.org:8002: /me chuckles
[11:04 AM] Andrew Hellershanks: :)
[11:04 AM] Lyr.Lobo @cc.opensimulator.org:8002: WE greeted Spring for you...early
[11:08 AM] Andrew Hellershanks: There were a number of commits this past week.
[11:08 AM] Lyr.Lobo @cc.opensimulator.org:8002: /me eyes the button
[11:09 AM] Vincent.Sylvester @hg.zetaworlds.com: Fun changes to webrtc still
[11:09 AM] Ubit Umarov: yeah a few changes
[11:10 AM] Ubit Umarov: i have it working on my test grid, either cong on a region as if standalone ( region talks directly to janus) or via grid
[11:10 AM] Ubit Umarov: parcels crossing a bit more resilient also
[11:11 AM] Ubit Umarov: but seems grid kind of config seems to fail badly to vicent
[11:11 AM] Ubit Umarov: still not clear why
[11:12 AM] Vincent.Sylvester @hg.zetaworlds.com: Did a test earlier moving entire blocks of code around to force load some stuff, but still no dice
[11:12 AM] Ubit Umarov: note that only parcels voice kinda works and only one janus gateway... other options are jsut broken
[11:13 AM] Vincent.Sylvester @hg.zetaworlds.com: It's almost as dysfunctional as Win11 has been lately xD
[11:13 AM] Andrew Hellershanks: hehe
[11:14 AM] Vincent.Sylvester @hg.zetaworlds.com: There is some architectural annoyances in the module design that bug the hell out of me, it's not how most things are designed. To put both the service and module code into one is a nightmare to debug
[11:14 AM] Vincent.Sylvester @hg.zetaworlds.com: Great re-use of code, annoying to figure out bugs with
[11:15 AM] Vincent.Sylvester @hg.zetaworlds.com: Tempted to split those things into proper parts and only load the things actually required with each type of setup to be able to debug what's going on
[11:15 AM] Vincent.Sylvester @hg.zetaworlds.com: Gonna become more important down the line as more features get implemented and the complexity goes up
[11:17 AM] Ubit Umarov: fs beta does include the fix for stun servers, guess cool vl also
[11:18 AM] Vincent.Sylvester @hg.zetaworlds.com: TJ reported changes to world map are coming to load more regions at once. Had a look at server side and tested a viewer with that change. Appears to be working fine. Internal anti-spam measures seem to only trigger past 150 requests while the change will do 64 at a time
[11:19 AM] Vincent.Sylvester @hg.zetaworlds.com: Suppose the idea is to load the world map a bit faster
[11:19 AM] Ubit Umarov: fs loves to flood servers
[11:19 AM] Vincent.Sylvester @hg.zetaworlds.com: Not sure why, never seen it load slow even with thousands of tiles to load
[11:24 AM] Vincent.Sylvester @hg.zetaworlds.com: It seems the memory increase from the http changes has somewhat dissipated somehow. Did some restarts today and the usage level is back to the usual amount. Though that possibly could be dotnet updates as well... if Microslop actually wrote some good code for a change
[11:25 AM] Vincent.Sylvester @hg.zetaworlds.com: I looked at the libgdiplus bug report again and it appears Canonical has re-assigned the maintainer for it to another team, but still no fix in sight and for the latest 26 version they are still building the broken version
[11:26 AM] Vincent.Sylvester @hg.zetaworlds.com: There is something bugged in pango and the way libgdi talks to it that causes pango to report font metadata, but no actual object when you request the font glyphs or something
[11:26 AM] Vincent.Sylvester @hg.zetaworlds.com: Trying to debug pango though... pain and a half
[11:27 AM] Vincent.Sylvester @hg.zetaworlds.com: May have to build libgdi exclusively with cairo, which works, but is said to be less accurate, whatever that means
[11:27 AM] Vincent.Sylvester @hg.zetaworlds.com: Thankfully this only manifests in system fonts of specific types not rendering while the standard fonts shipped with it are fine
[11:28 AM] Vincent.Sylvester @hg.zetaworlds.com: I did add some instructions to the wiki on how to get the good version of it and the dependency required directly rather than relying on the distro to provide broken versions
[11:32 AM] Ubit Umarov: on the http i did went back to that avoidas allocation of large buffers, adding timeouts in a dif way
[11:33 AM] Ubit Umarov: for webrtc ( possible others) i added option to set the size of the temporay buffer, like 1MB on one point
[11:33 AM] Ubit Umarov: doens form ms 2GB and my default limit of 256MB
[11:34 AM] Ubit Umarov: down...
[11:35 AM] Vincent.Sylvester @hg.zetaworlds.com: Memory doesn't grow on trees after all
[11:36 AM] Vincent.Sylvester @hg.zetaworlds.com: Last commit fixed a problem with texture entry encoding after the json changes. Pain trying to find the interactions on that
[11:37 AM] Vincent.Sylvester @hg.zetaworlds.com: Knock on wood that's the only thing
[11:38 AM] Ubit Umarov: ( was more on BP materilas encoding )
[11:38 AM] Ubit Umarov: materials even
[11:38 AM] Cuga.Rajal @rajal.org:9000: Any developments on that code to reduce DB writes?
[11:39 AM] Vincent.Sylvester @hg.zetaworlds.com: Most of that is in core. The only things I got left on top of that are the routine changes for primitems, because really that needs a more direct api
[11:39 AM] Vincent.Sylvester @hg.zetaworlds.com: Doing a read and compare is kind of a bandaid rather than a fix
[11:39 AM] Vincent.Sylvester @hg.zetaworlds.com: The bug with the linking being desynced is also still open, haven't had much time digging into that and it's a brain teaser for sure
[11:40 AM] Cuga.Rajal @rajal.org:9000: Yeah the primitems and linking/delinking is a brain exploder
[11:42 AM] Vincent.Sylvester @hg.zetaworlds.com: Has been oddly inconsistent even though it should be fairly reproducible, not sure what's going on with the async tasks run there. Looking at the db patch there are a few more things in the sop that I put some additional checks in that might need merging to core at some point
[11:42 AM] Vincent.Sylvester @hg.zetaworlds.com: A direct delete prim item would be useful in order to not have to read it all and compare just to remove a single item. I need to write that in there at some point
[11:44 AM] Cuga.Rajal @rajal.org:9000: But other db optimizations have been pushed to trunk. glad to hear
[11:45 AM] Vincent.Sylvester @hg.zetaworlds.com: Quite a few things on detecting inventory changes and sop changes are in already yes, which should reduce writes a little bit already
[11:46 AM] Vincent.Sylvester @hg.zetaworlds.com: With all the automatic stuff dotnet does in the background is kinda surprising it even triggers a set var to same value as a change, but then again I have learned not to trust it as far as I can throw it and my computer is pretty heavy xD
[11:47 AM] Vincent.Sylvester @hg.zetaworlds.com: Not sure that's what they meant with zero trust xD
[11:51 AM] Vincent.Sylvester @hg.zetaworlds.com: Had a strange thing happen a couple days ago. After simulator reboot items were missing, but only specific ones containing a script and animation. Either this was just bad shift copy or some other bug being masked by cache. Hasn't happened since thankfully
[11:51 AM] Vincent.Sylvester @hg.zetaworlds.com: Shift copy can fail on packet loss, but not at this scale I don't think
[11:51 AM] Vincent.Sylvester @hg.zetaworlds.com: Data was still on scene as the oars still had everything so must be something on db end
[11:52 AM] Vincent.Sylvester @hg.zetaworlds.com: I attract the weirdest bugs I swear
[11:52 AM] Vincent.Sylvester @hg.zetaworlds.com: why me ;_;
[11:52 AM] Andrew Hellershanks: :)
[11:52 AM] Ubit Umarov: mb you should get some phosphor piles to your cpus
[11:52 AM] Andrew Hellershanks: It keeps things interesting
[11:53 AM] Ubit Umarov: well or ram
[11:54 AM] Vincent.Sylvester @hg.zetaworlds.com: Already sacrificed a goat... oh you mean as in hardware xD
[11:55 AM] Ubit Umarov: No No No you will not sacrify our Dino!
[11:55 AM] Cuga.Rajal @rajal.org:9000: Which distro you using?
[11:56 AM] Cuga.Rajal @rajal.org:9000: or are you talkin Win
[11:56 AM] Vincent.Sylvester @hg.zetaworlds.com: This happened on ubuntu
[11:58 AM] Cuga.Rajal @rajal.org:9000: Hard to prove the distros make a difference, but I've had nothing but trouble with Ubuntu, a recent update clobbered and killed the EFI partition. I switched to Debian 13
[11:58 AM] Vincent.Sylvester @hg.zetaworlds.com: Way Canonical is currently messing things up switching is definitely on my mind
[12:01 PM] Ubit Umarov: ok ppl RL calls
[12:01 PM] Lyr.Lobo @cc.opensimulator.org:8002: Thank you for a great meeting. I must dash, first day back to classes. *waves*
[12:01 PM] Ubit Umarov: any other issue?
[12:01 PM] Andrew Hellershanks: ok, Lyr. tc. Thanks for dropping in.
[12:02 PM] Andrew Hellershanks: I'm interested in any work on libgdi as it affects another project I'm using.
[12:02 PM] Ubit Umarov: ok see you all next week if not before :)
[12:02 PM] Cuga.Rajal @rajal.org:9000: I'm going to convert from standalone to grid this week, and major gotchas I should know about? I think I know what to do on the config end.
[12:03 PM] Cuga.Rajal @rajal.org:9000: any* major gotchas
[12:03 PM] Andrew Hellershanks: Hope the upcoming time change in EU won't cause any problems for anyone.
[12:04 PM] Andrew Hellershanks: Thank you all for coming. See you again next week.
[12:04 PM] Vincent.Sylvester @hg.zetaworlds.com: If you have been using mysql db then probably not, perhaps some table name case might trip up, but usually that should work seamlessly. Last time I did that was years go though. Backup first and then work on a copy if you want to be sure
[12:04 PM] Cuga.Rajal @rajal.org:9000: Good tip thanks
[12:04 PM] Vincent.Sylvester @hg.zetaworlds.com: Andrew I'll keep you updated on libgdi via discord if I hear anything from Canonical or winehq
[12:05 PM] Vincent.Sylvester @hg.zetaworlds.com: I somewhat doubt either is going to fix it though
[12:05 PM] Andrew Hellershanks: Vincent, ok. ty
[12:05 PM] Andrew Hellershanks: I haven't been able to figure out how to get a usable backtrace when a program using libgdi dies. Makes it hard to pin down where in libgdi a problem exists.
[12:06 PM] Vincent.Sylvester @hg.zetaworlds.com: It's a problem in pango perhaps. It seems all the calls for "what is font called" and such return fine, but actually get the "font" object so to speak returns null
[12:06 PM] Vincent.Sylvester @hg.zetaworlds.com: I suspect pango does actually have the font loaded, but whatever provides data to outbound isn't converting it correctly or something along those lines
[12:07 PM] Vincent.Sylvester @hg.zetaworlds.com: The suspect commit in libgdi was a change to pango min version loading newer versions now which leads me to think that it must be pango changes, but digging into that history is pain
[12:08 PM] Andrew Hellershanks: Seems reasonable. There was also a known bug in one of the routines regarding a cursor related function where the function is passed an invalid value.
[12:09 PM] Vincent.Sylvester @hg.zetaworlds.com: Would be an option to bug the pango maintainers to have a look at the test that is failing for libgdi
[12:09 PM] Vincent.Sylvester @hg.zetaworlds.com: Maybe they know something
[12:09 PM] Andrew Hellershanks: Is there a way to build libgdi such that it keeps its symbol table that would be useful for debugging?
[12:10 PM] Vincent.Sylvester @hg.zetaworlds.com: Probably, not sure how. I don't do much C work
[12:10 PM] Vincent.Sylvester @hg.zetaworlds.com: Probably compile flag or mode or something
[12:10 PM] Andrew Hellershanks: I read something that said that msbuild now strips symbol info by default.
[12:11 PM] Vincent.Sylvester @hg.zetaworlds.com: Something with configure autobuild or even cmake perhaps, I mostly followed libgdi compile instructions as written which I guess should compile in debug
[12:12 PM] Vincent.Sylvester @hg.zetaworlds.com: If not then it must be some flag or option to pass to get it to
[12:12 PM] Andrew Hellershanks: I did the same but it still didn't seem to help .
[12:13 PM] Vincent.Sylvester @hg.zetaworlds.com: C is a bit different when it comes to debugging so no idea, been too long that I really messed with that, might not have enough things to really debug at all
[12:13 PM] Vincent.Sylvester @hg.zetaworlds.com: C is odd
[12:13 PM] Andrew Hellershanks: Part of the problem is that it is a library rather than something in the main program.
[12:15 PM] Vincent.Sylvester @hg.zetaworlds.com: Might be best just to write a small test case and file it as bug report, might attract someone going "well actually you are doing it wrong" in best case heh
[12:15 PM] Andrew Hellershanks: Yes, but I don't know the library that well or what it is that is triggering the problem so writing a test case would be a shot in the dark.
[12:16 PM] Vincent.Sylvester @hg.zetaworlds.com: Can't fix everything, sometimes gotta hand it off and have someone else wreck their brain over it
[12:17 PM] Andrew Hellershanks: True.
[12:18 PM] Andrew Hellershanks: Gotta get back to work. ttyl
