Time Nick Message 08:31 sfan5 getinfo is already gone 08:33 sfan5 I merely documented the reality 10:44 Pexin suggest giving #11939 the bugfix and trivial tags. bugfix because currently touching_ground while bouncing is very inconsistent. 10:44 ShadowBot https://github.com/minetest/minetest/issues/11939 -- Restore bouncy behavior by pecksin 15:13 rubenwardy Pexin: movement changes are rarely trivial, even if the code is small 15:27 erlehmann i do not like that the PR touches two things 15:27 erlehmann but maybe it has to? 16:11 sfan5 it appears I'm wrong about getinfo 16:55 Pexin rubenwardy: i understand. i made sure to restrict the effect to only bouncy floor at the moment of impact, but am aware i dont know the extent of what touching_ground affects 17:37 MTDiscord sfan5: huh? WDYM? 17:38 MTDiscord builtin uses debug.getinfo 17:42 erlehmann Pexin have you tried out some complex games with the ground touching thing? 17:43 erlehmann or just devtest? 17:45 sfan5 @luatic only the CSM code undefines it after usage, so it is in fact available to mods 17:48 MTDiscord sfan5: yeah, just realized I was half-right 17:53 Pexin erlehmann: i've been using my patch with linux forks server 17:53 Pexin it's "pretty complex" 17:54 Pexin no differences noted with swimming or "non jumpable" surfaces 17:55 erlehmann whhat is “linux forks server”? 17:55 Pexin "this server has a lot of trains" 17:55 Pexin and custom technic etc etc 17:56 Pexin also tested with a local singleplayer using a different trampoline mod 17:57 Pexin the reason it touches 2 files is because it does 2 related things. the ground collision, and new "sneak to reduce height" functionality 17:58 erlehmann in what way are they related? 17:58 erlehmann could it be separate PRs? 17:58 erlehmann i mean it has to be separate commits ofc 17:58 erlehmann so one can be reverted easily if it is wrong, while keeping the other 17:59 Pexin the reduce height thing depends on ground collision, and the ground collision alone would result in trampolines that can accumulate, but not decrease height 18:01 Pexin I mean if someone makes a trampoline that's above 100% bouncy that's a deliberate modder choice 18:02 erlehmann ok so when exactly is the height reduced? 18:02 erlehmann i worry about this maybe messing up parkour 18:02 Pexin at the moment of impact with bouncy floor (floor only, not walls or ceiling) if sneak is held at that moment 18:03 erlehmann so that means it is impossible to trap someone in a bouncy prison? 18:03 erlehmann or, otherwise, if someone sneaks while going down, they might get reduced height? 18:03 Pexin not sure what you mean by that, but if bouncy is at most 100% they can always force themselves to settle 18:04 Pexin if sneak is held at the moment of impact, the recoil is reduced from what it would normally be 18:05 Pexin if sneak affects normal falling, that's outside this scope, and it would surprise me 18:05 Pexin nothing to do with this pr 18:05 erlehmann so can this feature of sneak be turned off? 18:05 erlehmann i mean i find it useful 18:05 Pexin can bouncy be turned off? 18:06 erlehmann well that is clear, it can 18:06 erlehmann but i mean the bounce reduction 18:06 Pexin not in the current state of the PR but i doubt that would be useful 18:06 erlehmann like what if as a game you want to make it impossible for the player to bounce less 18:06 Pexin hm 18:07 erlehmann or what if a game or mod uses sneak for something else 18:07 erlehmann for example, there is a mod by fleckenstein where sneak shows who is currently connected 18:07 Pexin this would be entirely new behavior for trampolines, I was trying to restore the usable functionality of pre-5.3 bouncy, which requires new reduction feature 18:07 Pexin hm 18:07 erlehmann i do not understand, how does it require it? 18:07 erlehmann i am a bit sleepy though 18:07 Pexin otherwise there's no way to reduce bounce 18:08 erlehmann and there was before, i assume? 18:08 Pexin old behavior was weird in that it decayed noticably even at 100 18:08 Pexin which seems wrong, and 5.3 changed it irrevocably 18:09 erlehmann i am not saying that the sneak thing is a bad feature per se 18:09 erlehmann i think it is nice 18:09 erlehmann i just try to figure out where it might lead to unexpected things 18:09 erlehmann i would say, ask a few more people 18:09 Pexin if a mod uses sneak to display info, i would hope it doesn't like.. create a popup window 18:10 erlehmann it is a hud overlay in this case 18:10 erlehmann what would be bad about the popup? does sneak get stuck? 18:10 Pexin if it doesn't _interfere_ with normal sneak behavior it shouldn't matter I'd think 18:10 erlehmann since i play mostly on servers where ppl are allowed to use cheat clients and we take that into account party during development, i think it is not too bad for whatever i am doing as a player or dev 18:11 Pexin well if it creates a popup formspec then it totally interferes with gameplay 18:11 erlehmann in fact, i like the feature 18:11 erlehmann nah look for mods by fleckenstein 18:11 erlehmann it is pretty useful i think! 18:30 sfan5 merging game#2871, game#2751 in 8m 18:30 ShadowBot https://github.com/minetest/minetest_game/issues/2871 -- Allow to set use_texture_alpha when registering a door by Emojigit 18:30 ShadowBot https://github.com/minetest/minetest_game/issues/2751 -- player_api: set_texture() to set single textures by Zughy 18:32 MTDiscord sfan5: wait a minute 18:32 MTDiscord with 2751 18:33 MTDiscord sfan5: the given implementation looks incorrect to me 18:34 MTDiscord it doesn't update the internal player_textures table 18:34 MTDiscord so it creates an inconsistency: player properties are changed, but player_api.get_animation will still happily return the outdated textures 18:34 sfan5 oh indeed 18:35 sfan5 if you want to rebase game#2745 and add it in there instead we can merge that 18:35 ShadowBot https://github.com/minetest/minetest_game/issues/2745 -- Redo player_api by appgurueu 18:35 MTDiscord why does it need a rebase? 18:35 MTDiscord did anything touch player_api? 18:36 sfan5 github says so if you select the wrong merge type (which I did) 18:36 MTDiscord hmmkay 18:44 erlehmann > it creates an inconsistency: player properties are changed, but player_api.get_animation will still happily return the outdated textures 18:44 erlehmann that sounds like a thing that a test case or an assert could catch, no? 18:57 sfan5 do you think MTG has tests 19:09 MTDiscord minetest.after's job cancelling leaks memory as it merely marks the job as cancelled, still keeping a reference to the now unneeded arguments around until the job expires 19:15 sfan5 submit a fix? 19:22 MTDiscord sfan5: it isn't trivial 19:22 MTDiscord modlib has some datastructures for it, but it currently uses the same lazy implementation for :cancel() 19:23 sfan5 removing from the list is not trivial 19:23 MTDiscord the problem is that removing the job from the list would be linear time 19:23 sfan5 doing .args = {} is trivial 19:23 MTDiscord yes 19:23 MTDiscord still a memleak, albeit an arguably much smaller - and most likely negligible - one 19:24 MTDiscord the correct data structure here is what I call a hash heap though :P 20:08 MTDiscord removing jobs from the list isn't that hard, i have a working after implementation for that 20:09 appguru Are you doing the swap thingy? Always forget that exists :P 20:09 MTDiscord besides memory leak it has another problem, if you spawn and cancel a lot of jobs, you still will iterate over them at ever globalstep, globalstep becomes slow 20:09 appguru yeah 20:09 appguru the entire list is the wrong approach 20:09 appguru should be using a heap 20:10 MTDiscord yeah i use swap, builtin after also uses swap, just not for cancel 20:11 sfan5 appguru: do I see correctly that model textures take precedence and effectively make any mod unable to change the player textures? 20:12 MTDiscord let me fix that 20:12 MTDiscord dumb copy-paste from set_textures 20:12 sfan5 well your PR consistently does that, I wonder why 20:14 MTDiscord yeah, gotta fix set_textures too 20:15 MTDiscord https://github.com/minetest/minetest_game/pull/2745/commits/1de58d0011c065b8175a062a3d8acff72d83d9be