Time Nick Message 01:44 v-rob As requested, I got some of my Unicode stuff to compile and put it in PR form: #12066 01:44 ShadowBot https://github.com/minetest/minetest/issues/12066 -- Add UniStr class for all Unicode string handling by v-rob 04:48 MTDiscord I just want to put this idea out there... I think that textures should be able to be defined for nodes, such that you see a different texture based on your view angle. Maybe it selects the texture from an animation tile. This would allow for example, shiny metal, which the reflection areas change based on your view angle 04:49 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/941556520276074566/unknown.png 04:52 erlehmann Mister-E that's a cool idea that would work even without shaders. have you tried implementing it as a CSM? 04:55 MTDiscord I have no experience with even the csm api, and it seems like smth that would require engine modification. No, I was just putting the idea out there to the engine devs 04:55 MTDiscord a new node drawtype I guess 04:56 MTDiscord or maybe a new tile animation type, and its used for objects too? that would be good perhaps 10:00 MTDiscord Angle based tiles sounds really useful 10:01 MTDiscord But I think you may underestimate how many are needed for a given effect 10:04 erlehmann I have no experience with even the csm api, and it seems like smth that would require engine modification. 10:04 erlehmann hmmm 10:04 erlehmann not sure 10:04 erlehmann maybe ws or df can do it? 10:05 erlehmann well, fleck stopped doing minetest stuff :/ 12:54 MTDiscord Angle-based tiles sound like you'd have to do a lot of scene manipulations per-frame to swap out the tiles, and ultimately for most users, that might actually perform notably worse than simple shaders... 12:55 MTDiscord If you're looking for something a bit specific, like glare or something, wouldn't something like an envmap work for that, while still letting us use the old fixed-function rendering pipe? 12:57 MTDiscord Shiny metal should use specular lighting (shader change). This is in fact the way simpler way to go. If a "pixelated" effect is intended, that complicates mattesr a slight bit, but is still preferable over spamming textures. 13:10 erlehmann i would turn off this effect anyway 13:10 erlehmann i have an attention deficit, shiny stuff distracts me from gameplay 17:14 erlehmann why was this closed? https://github.com/minetest/minetest/issues/9352 17:15 MTDiscord Look at the close message, which links https://github.com/minetest/minetest/pull/10887 17:17 erlehmann this only mentions the leaving case. is there no corresponding race for the join? 17:18 sfan5 instead of asking "why was closed??" why not explain the problem you are facing 17:19 erlehmann ok. i think that retrieving anything about players is racy in general, because the list of connected players can include players that are not yet there or have already left. 17:20 erlehmann in this issue someone triggered the join case using a python script i think. not usere. https://git.minetest.land/Mineclonia/Mineclonia/issues/269 17:20 erlehmann not sure 17:21 erlehmann does the engine guarantee that minetest.on_joinplayer() has run when the player appears in the result of minetest.get_connected_players() maybe? i do not think so. 17:23 sfan5 I am pretty sure it does 17:24 sfan5 but if there's a packet that can confuse the server's state machine then that should be looked at first 17:24 erlehmann https://gist.github.com/corarona/b9c726527a0ddb063ccb8b8b7e2ffc65 17:24 erlehmann cora says this can be used 17:25 erlehmann with https://github.com/Lejo1/mtmodule 17:25 erlehmann put in mtbot/protocol.py 17:26 erlehmann “a packet that can confuse the server's state machine” sounds alarming. is the state machine documented anywhere but in the code directly? 17:27 sfan5 does a code comment count 17:27 erlehmann not really, more like graphviz or plantuml 17:27 erlehmann i.e. the high level overview how it *should* go 17:27 erlehmann regardless of whatever the code does (bc code often does weird stuff) 17:27 sfan5 the comment in question is an ascii chart 17:27 erlehmann oh ok that counts 17:28 sfan5 src/clientiface.h 17:28 erlehmann thx 17:28 erlehmann oh i have seen this before