Time  Nick      Message
04:56 ANAND     I'm trying to implement wielditem checking in one single place that takes care of everything.
05:00 ANAND     Apart from wield *index* changes, of courese
05:00 ANAND     course*
05:01 ANAND     Is there possibly a better way to do this than to store the "old" wielditem and keep checking in every ServerEnvironment / PlayerSAO step or smth?
06:28 ANAND     Oh, we're in feature freeze now. :O
08:40 sfan5     ANAND: yes, check when inventory events come in
08:48 sfan5     the other PR has an implementation in fact https://github.com/minetest/minetest/pull/9862/files/7489977e577d9b4e679aff5fe3ee82e9ac070bc0..5cf4828971bc86ff9c0a3e8c0d852ebf6e17e22b
09:04 ANAND     Yeah, I went over the changes this morning
09:04 ANAND     But that PR is incomplete too
09:05 ANAND     It manually checks for wielditem changes in individual places like on_place, etc.
09:05 ANAND     If another way to modify the inventory is added, the on_(un)wield callbacks would break
09:06 ANAND     I'd ideally like to implement wielditem checking in one single place where all inventory changes pass through
09:09 sfan5     lua inventory actions do not trigger any callbacks, which is sort of an issue currently
09:11 ANAND     Yeah
09:11 ANAND     I really wish they do
09:12 ANAND     Oh wow, I'm only handling key release events for mouse buttons! Haha!
09:13 ANAND     Both issues caught by Krock are related to key release events. Ty, Krock :P
09:19 ANAND     Yeah, that fixes it. lol...
09:21 ANAND     wasKeyDown is really weird. It's used as a drop-in replacement for a proper event-callback system.
09:21 ANAND     i.e. It retains its state for only one call
09:22 ANAND     Sorta like "eating" the event
10:12 ANAND     #7924 is ready for review. I've fixed all the things (TM).
10:12 ShadowBot https://github.com/minetest/minetest/issues/7924 -- Allow binding dig, place actions to keys; remove LMB/RMB hardcoding by ClobberXD
11:34 mntmn     opinions? https://github.com/minetest/minetest/pull/10036#discussion_r440105528
11:39 erlehmann i have tested https://github.com/minetest/minetest/pull/10036 on my hardware – Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller (rev 03) – and a scene that took between 730ms and 745ms to render without the fix from mntmn took about the same time with the fix from mntmn
11:40 PGimeno   mntmn: isn't that how it's done currently?
11:41 mntmn     PGimeno: not sure, the side effects that i saw here with depth testing not working correctly suggested that the nodes are drawn in a mixed fashion
12:01 sfan5     https://github.com/minetest/minetest/blob/master/src/client/clientmap.cpp#L277
12:01 sfan5     ClientMap::render has a transparent pass but it's apparently only used for liquids
13:10 PGimeno   only for liquids? that somewhat explains why liquid seen through glass becomes invisible sometimes
13:13 PGimeno   LuaJIT 2.1 appears to be stable on Android again: https://love2d.org/forums/viewtopic.php?f=3&t=86051&p=234219#p234219 (and the next message)
13:23 ANAND     Oh nice, #9961 has been merged already :)
13:23 ShadowBot https://github.com/minetest/minetest/issues/9961 -- Server pushing media at runtime by sfan5
20:17 oil_boi   So I seem to have found the source of https://github.com/minetest/minetest/issues/10024 krock. client->network->enable_client_modding is disabled by default. Which brings up a whole slew of other issues because if this was disabled and I was able to trigger it by the method I showed you then this means the client is overriding the manual network setting to load a client mod when logging into a server
20:18 sfan5     then the other behaviour is the bug
20:18 sfan5     if you join a singleplayer game whose game has "enable_client_modding = true" in its config, the next time you join a server client modding will stay enabled
20:18 sfan5     does that sounds like what happened?
20:18 oil_boi   Absolutely
20:19 oil_boi   So I showed people how to trigger a bug to log into the server good
20:19 oil_boi   Also, why not enabled by default?
20:22 sfan5     because csm is unfinished and there's nothing useful for CSM in the default install (maybe the preview mod is even left enabled) probably
20:24 oil_boi   But it is ultra extremely useful for modders that want to use the processing power of clients to distribute game calculations along with negate masses of latency away from the server
20:27 sfan5     it's obvious that CSM itself is useful, that's not what I meant
20:31 oil_boi   sfan5, there https://github.com/minetest/minetest/issues/10038
20:38 oil_boi   I am going to learn the main menu api and create a new tab just for it
20:39 sfan5     you can add a button that enables the setting there too
20:42 oil_boi   This is my disgusting mock up https://i.imgur.com/Cn8tepU.png
20:44 sfan5     for the mod list i'd make it function exactly like the normal list in the world menu
20:44 sfan5     world configure menu*
20:45 oil_boi   I'll take a look at that, this main menu api seems pretty intuitive
20:58 oil_boi   So we have 3 options here. One, I shove this into content and make that a padding nightmare. Two, I repad the settings menu since https://i.imgur.com/OwlqbbJ.png there is a lot of wasted space, or three use allll that empty space at the top for a new "client mods" tab which controls client modding
21:00 sfan5     add a button somewhere that opens a submenu?
21:00 Wuzzy     hello
21:00 Wuzzy     have you gotten my script link?
21:00 oil_boi   Hi wuzzy
21:00 Wuzzy     (translation thing)
21:01 sfan5     seen it? yes tried it? no
21:01 oil_boi   Oh sfan5 you're very right, right below the center tab of the settings menu, between "all settings" and "change keys" there is room for a third button!
21:01 Wuzzy     ok
21:02 oil_boi   Is there a warning forum post that a core dev posted about csm being experimental?
21:05 Wuzzy     i think not
21:16 oil_boi   Before I move any further hows: https://i.imgur.com/dadjhRo.png
21:34 Wuzzy     Unmerged bugfix PR with 1 approval: #9486
21:34 ShadowBot https://github.com/minetest/minetest/issues/9486 -- Play place_failed sound if node is occupied or attaching node fails by Wuzzy2
21:35 oil_boi   Oooo wuzzy that would be noice
21:36 Wuzzy     wish me luck xd
21:36 oil_boi   GL GL
22:03 Wuzzy     dumb question and probably off-topic but anyone has an idea why MCL2 biomes dont work in 5.3.0-dev but work in 5.2.0? ?
22:03 Wuzzy     was there any changes to biomes that could have triggered this?
22:03 sfan5     explain "don't work"
22:03 Wuzzy     all-stone world in non-v6
22:04 Wuzzy     I had that for months in -dev now, but decided to ignore it. ? but now we're in freeze, so i am forced to care ?
22:08 sfan5     >mg_flags = caves, nodungeons, light, decorations, nobiomes
22:09 erlehmann sfan5, you wrote here that the code can be put behind an option if it is bad for performance, but so far no bad performance effect has been demonstrated, am i wrong? https://github.com/minetest/minetest/pull/10036#issuecomment-644330088
22:09 sfan5     erlehmann: correct
22:10 sfan5     Wuzzy: there is no biomes checkbox in the menu so perhaps the menu code is somehow at fault?
22:10 oil_boi   Lone wolf suggested this could be injected into the content page and it actually works
22:11 sfan5     not sure if the content page is fitting tbh
22:11 erlehmann sfan5, would hecktest now have to demonstrate that the GL_ALPHA_TEST has different performance characteristics than the discard in the shader or how to proceed?
22:11 Wuzzy     sfan5: MTG works
22:11 Wuzzy     I made sure the biomes flag is not touched in the world menu
22:11 sfan5     well its disabled when I create an mcl world
22:11 sfan5     so somehow...
22:12 Wuzzy     oh? wtf
22:12 sfan5     erlehmann: if there are possible performance problems with the discard shader I suggest putting it behind a check
22:12 sfan5     because that can't hurt
22:13 Wuzzy     sfan5: /set mg_flags yields biomes flag for me, not nobiomes. you?
22:13 Wuzzy     sfan5: fyi: the biomes checkbox does not exist in the menu, ture. but this just means whatever state the biomes flag is in the settings, it is preserved
22:14 sfan5     Wuzzy: says everything is enabled
22:14 Wuzzy     the problem can't be the world creation menu, otherwise it would break MTG too :/
22:14 Wuzzy     maybe the bug is in MCL2, but for *some* reason it never manifested itself in 5.2.0 ?
22:15 Wuzzy     any idea how i could screw up biomes gen in a game? ?
22:15 sfan5     ACTION[Server]: singleplayer tried to execute print(core.get_mapgen_param("mg_flags"))
22:15 sfan5     caves,nodungeons,decorations,light
22:15 sfan5     ^ this is the proper way to ask for that setting btw
22:15 sfan5     and as I've just noticed biomes is missing which might be related
22:16 Wuzzy     O_O
22:16 Wuzzy     weird
22:16 Wuzzy     lets check if other games are broken as well
22:17 sfan5     it's not weird at all considering ./mods/MAPGEN/mcl_mapgen_core/init.lua literally sets it like that
22:17 sfan5     does #9746 fix this perhaps?
22:17 ShadowBot https://github.com/minetest/minetest/issues/9746 -- Settings: Unwrap overcomplicated flags calculation by SmallJoker
22:17 Wuzzy     oh noes!!
22:17 Wuzzy     i didnt set biomes flag? hahahaha
22:19 Wuzzy     Yes, that was it! omgggg i feel ashamed now
22:19 Wuzzy     hmmm i should probably move to the new flag function after release. is better coding anyway
22:21 Wuzzy     I shall test #9746 now...
22:21 erlehmann sfan5, i prefer sane defaults, so the question is how to enable the conditional discard only on the vivante GPU in question
22:21 ShadowBot https://github.com/minetest/minetest/issues/9746 -- Settings: Unwrap overcomplicated flags calculation by SmallJoker
22:22 sfan5     Wuzzy: offtopic: you should change sprinting to use a particlespawner instead of calling add_particle multiple times, it also always spawns dirt particles which is easy to fix with the new `node` parameter
22:23 Wuzzy     will look at it, assuming i can attach spawner to player
22:23 sfan5     erlehmann: you can wire up a check for the GL_RENDERER string to the shader code to do a '#define DISCARD_WORKAROUND 1' (or 0) and then put the code into and #if #endif
22:23 sfan5     Wuzzy: that's possible too, but I'm just suggesting to replace multiple add_particle calls with a single particlespawner that spawns N particles
22:23 sfan5     (no idea how your code looks, just guessing)
22:24 Wuzzy     oh, thats obvious. yes that part of code is garbage then
22:32 PGimeno   "If you must use discard make sure that only the tris that need it are rendered with a shader containing it and, to minimise its effect on overall rendering performance, render your objects in the order: opaque, discard, blended."
22:33 PGimeno   that's from the link provided by hecks
22:41 oil_boi   The main menu has no access to the mods.conf file in clientmods?
22:44 oil_boi   Disregard that
22:45 Wuzzy     sfan5: no, #9746 does not fix that.
22:45 ShadowBot https://github.com/minetest/minetest/issues/9746 -- Settings: Unwrap overcomplicated flags calculation by SmallJoker
22:45 Wuzzy     but don't worry, when i explicitly add biomes to flag list, MCL2 works again.
22:46 Wuzzy     still it is a bit troublesome that minetest falls back to bad default behavior if i fail to provide all flags explicitly ?
22:48 sfan5     I don't know if "if the user doesn't specify, assume a flag to be disabled" really is bad
22:49 Wuzzy     indeed
22:50 Wuzzy     also, keep in mind it worked in 5.2.0. this might be a regression!
22:50 Wuzzy     biomes flag should default to ON for obvious reasons ?