Time  Nick       Message
02:05 Foz        I've been investigating a race condition bug. It seems like under some circumstances and LBM can run before all the entities in a map block have loaded. Does anyone know about this or how to fix it?
02:07 sfan5      I only know of this LBM-related crash: https://github.com/sfan5/minetest/commit/5b65da6761d1e2523138c3755ff807a1298e9b08
02:07 sfan5      not sure if that was ever fixed, but i've never seen it happen agian
02:07 sfan5      again*
02:17 Foz        I've been trying to fix this: https://github.com/minetest-mods/homedecor_modpack/issues/405
02:18 sfan5      are you sure LBM are related to this?
02:18 Foz        I discovered that itemframes and pedestals were creating extra entities because the entities were sometimes not loaded when an ABM that checks for them ran.
02:19 Foz        I changed the ABM to an LBM, yet the LBM is also sometimes running before all the entities are loaded, causing the same problem.
02:19 Foz        I need to find a way to ensure everything is loaded before running the check code.
02:20 Foz        but I kind of think that LBM's should provide that assurance
02:20 sfan5      entities are sort of a special case, wouldn't surprise me if it doesn't
02:21 sfan5      it doesn't suffice to just load a mapblock, the entities also need to transition into an "active" state, so they have a globally unique id in the current server state
02:21 sfan5      this is done seperately in the ServerThread loop so it may well not be done when LBMs run
02:22 Foz        interesting
02:23 Foz        do you have any idea how to trigger the check after entities are finished becoming "active"?
02:24 Foz        I have tried just using a minetest.after() but that has problems if the map block gets unloaded
02:25 sfan5      the only callbacks you get during entity activation are for each individual entity, nothing e.g. per-mapblock
02:25 sfan5      so no there's no obvious way to do trigger a check after a block finishes activating
02:25 sofar      me and est31 tested LBMs pretty well back when it was merged
02:28 Foz        I don't think this is a bug with LBMs as such.  I think the bug was the assumption that all the entities would be loaded when this particular LBM ran.  Unfortunately there is no guarantee of that.
02:30 sofar      I would actually expect LBMs to run before entities even
02:30 Foz        itemframes and pedestals display items with an entity.  If the entity gets deleted, say by running //clearobjects, the mod tries to test for that and replace it using an LBM, but this sometimes (through very rarely) fails.
02:31 Foz        As of yet, I haven't been able to think of any way around this problem
04:40 VanessaE   sofar: er...if map data has been loaded, but not the entities, then imho the block is not "loaded" yet.
04:40 VanessaE   at least for statically-stored entities.
04:41 VanessaE   maybe add a flag to force an LBM to only run after entities have been loaded?
04:42 VanessaE   Foz: the only solution I can see is to go back to using an ABM, written to clear out any extra entities it finds and re-setup just the one that ought to be there?.
04:43 VanessaE   (if there's more than one near a pedestal or frame)
06:41 sofar      there's something weird going on witn MT on win10 and 4K displays
06:41 sofar      first, the mouse doesn't want to center and so the screen is constantly drifting down and right
06:42 sofar      even though the mouse is not moving
06:42 sofar      and second, it's completely confused about the screen resolution
06:42 sofar      if I tell it to use 1920x1080 wxh, it's giving me more something like 2800x1550 or so
06:43 sofar      off by 30% or so
06:43 sofar      really bizarre that it can't even set it's own window width properly on startup
06:46 VanessaE   could it be some odd DPI-scaling issue?
06:46 sofar      possible, except MT still had the default 72dpi set, so you'd think it would make it smaller not larger
06:51 VanessaE   I was thinking more along the lines of window-manager-level (well, whatever Windows calls their equivalent)
06:51 VanessaE   is there some hint you have to set?
06:52 sofar      no idea, just using a stock build 0.4.16 from the main website, that's all
07:31 epoch      null pointer dereference in fairly fresh git clone build
07:32 epoch      (gdb) x/1i $eip
07:32 epoch      => 0x5c7dc9 <_Z24createOpenALSoundManagerP21SoundManagerSingletonP20OnDemandSoundFetcher+73>:mov    0x4(%ecx),%eax
07:33 epoch      git says already up to date so...
07:33 epoch      I'm not complaining because this would keep me from using minetest
07:33 epoch      just letting you know
07:34 epoch      and if I didn't say it here I know I'm too lazy to do a pull request atm.
07:34 epoch      game.cpp
07:35 epoch      oh good. easy way to skip around that code without commenting it out and rebuilding.
07:41 epoch      looks kind of like it is ignoring the enable_sound being set to false.
07:56 epoch      the GUI Engine tries to do createOpenALSoundManager too
07:56 epoch      but does not care about minetest.conf
08:27 epoch      should the config be checked in sound_openal.cpp's createOpenALSoundblahblah
08:27 epoch      ?
10:40 celeron55  sofar: windows has those weird scaling modes that it enables unless the program is somehow specially marked or calls some api
10:41 celeron55  you can manually disable them for each program shortcut in the compatibility options
10:42 celeron55  ...sometimes
11:49 sfan5      pushing trivial https://a.uguu.se/YYeYsKRsEHdd_EQCGD8n.txt in 5 minutes
11:55 nerzhul    nice sfan5
11:59 Krock      sfan5, you forgot to check whether "creative" is available
11:59 Krock      fireflies does not depend on that mod, but it will most likely work as long creative isn't removed from MTG by players
12:01 sfan5      > if not (creative and creative.is_enabled_for and creative.is_enabled_for(player_name)) then
12:01 sfan5      pretty bothersome if you ask me
12:02 Krock      well, creative.is_enabled_for is most likely available, so only `if not (creative and creative.is_enabled_for(player_name)) then` would be required
12:03 sfan5      existing code is inconsistent on this https://a.uguu.se/bYwqtsvQCv7E_a.html
12:05 Krock      they all include a check for the `creative` table
12:05 sfan5      anyway, fixed
12:05 Krock      hehe :) thanks
12:06 sfan5      some don't check for creative.is_enabled_for though
12:06 Krock      yes, creative itself for example
12:06 sfan5      ...and "map" and "binoculars"
12:07 Krock      oh. checking for the function. I see. Indeed, they don't check for that
17:01 rubenwardy Krock, sfan5: CLion messes everything up, think I'll go back to atom
17:02 Krock      no possibility to tweak the formatting rules?
17:03 rubenwardy there is, but it'll still cause formatting changes in weird places
17:31 nerzhul    rubenwardy: messed what ?
17:31 nerzhul    formatting is easy to customize in clion
17:31 rubenwardy #7099
17:31 ShadowBot  https://github.com/minetest/minetest/issues/7099 -- Add formspec theming using prepended strings by rubenwardy
17:31 rubenwardy not when it's as inconsistent as Minetest
17:31 rubenwardy it just changes random shit
17:31 nerzhul    if it does wrong formatting it can be due to wrong settings, mine respects mt rules :p
17:32 rubenwardy not wrong formatting
17:32 rubenwardy formatting on lines I don't touch
17:32 nerzhul    do you have up-to-date soft ?
17:32 rubenwardy probably
17:32 nerzhul    it doesn't on mine
17:32 nerzhul    except ending file
17:32 rubenwardy \o/
17:32 nerzhul    (on commit for this part)
17:47 rubenwardy Krock, sfan5: updated
17:49 Krock      how did test_config.h get in there?
17:49 rubenwardy because the gitignore probably sucks
17:50 rubenwardy removed
17:53 Krock      Code looks good. Testing now
17:55 Krock      uhm.. it's showing a merge conflict here.. trying to apply the diff
17:57 nerzhul    test_config.h comes from unittests
17:57 nerzhul    it's in the .gitignore, or maybe i missed it
18:01 rubenwardy Krock, rebased and force pushed
18:01 rubenwardy I suggest using git pull instead of applying the diff, hwoever
18:02 rubenwardy git checkout HEAD~200;  git pull https://github.com/rubenwardy/minetest formspecprepend
18:02 rubenwardy kinda wish there was a command to replicate a remote branch
18:02 rubenwardy without doing that
18:02 rubenwardy like, it would automatically create the branch and switch to it
18:03 Krock      thanks, the new patch worked
18:04 Krock      some kind of "shallow" checkout is possible too, so checkout HEAD~200 won't be needed if it succeeds
18:13 Krock      https://pastebin.com/raw/hmXPCEqB but works
18:48 rubenwardy Krock: can I just disable that warning
18:48 rubenwardy it's BS
18:48 rubenwardy maybe another PR
18:54 p_gimeno   rubenwardy: https://gist.github.com/pgimeno/4e1c1a8e78da1286b57bfca3a8b66ff7 - no need to rewind HEAD
18:54 rubenwardy oh cool
18:55 rubenwardy didn't know checkout could do that
18:55 rubenwardy Krock, updated
18:56 rubenwardy really need to get hub installed, anyway
18:56 rubenwardy that would probably help a bit
18:56 p_gimeno   you can just check it out without creating a local branch, if all you need is take a look: git checkout rubenwardy/formspecprepend, but you get a scary warning message (just git checkout master "solves" it)
18:58 rubenwardy rthanks
19:55 nerzhul    yes, checkout the remote as detached is better when you don't need to perform local branch modifications
19:56 nerzhul    my local repository has 3 remotes, upstream (mt), origin (my fork) and weblate. To rebase i always perform git fetch upstream && git rebase upstream/master for example :)
20:36 rubenwardy updated #6396
20:36 ShadowBot  https://github.com/minetest/minetest/issues/6396 -- Add reasons to on_dieplayer and on_hpchange by rubenwardy
20:48 nerzhul    rubenwardy wants travis to burn :D
20:48 rubenwardy soz
21:03 rubenwardy sfan5, updated #7099, but without u32 -> u16 change (see thread)
21:04 ShadowBot  https://github.com/minetest/minetest/issues/7099 -- Add formspec theming using prepended strings by rubenwardy
22:15 paramat    will merge #7164 in 2 hrs
22:15 ShadowBot  https://github.com/minetest/minetest/issues/7164 -- Mapgen: Remove unused mgv7 code and some biometypes by paramat
22:18 nerzhul    paramat 7164 is okay for me
22:18 paramat    thanks for looking
22:36 rubenwardy sfan5, updated #7099
22:36 ShadowBot  https://github.com/minetest/minetest/issues/7099 -- Add formspec theming using prepended strings by rubenwardy
22:36 rubenwardy thanks for pointer
22:36 rubenwardy *thanks for the pointer
22:40 paramat    mgcarpathian mapgen loop can be optimised, will work on that, and some of it is a mess, i won't alter terrain though
23:26 Warr1024   "*thanks for the pointer" <-- shouldn't that be "thanks for the *pointer" ?
23:26 rubenwardy lol
23:26 rubenwardy probably
23:55 rubenwardy nerzhul, what's the name of the PR to add lua unit tests?
23:56 rubenwardy I believe it was closed
23:56 rubenwardy *merged
23:56 rubenwardy aw, ffss
23:57 rubenwardy please include stuff a PR has in its title
23:57 rubenwardy #7131
23:57 ShadowBot  https://github.com/minetest/minetest/issues/7131 -- Server: delegate mod management & config to ServerModConfiguration by nerzhul