Time Nick Message 08:57 sfan5 merging #17614, #17615, #17602, #17595 in 15m 08:57 ShadowBot https://github.com/luanti-org/luanti/issues/17614 -- Verify paths passed to `load_schematic` with `CHECK_SECURE_PATH` by user333dev 08:57 ShadowBot https://github.com/luanti-org/luanti/issues/17615 -- [don't squash] Some refactoring + Lua CLI proof of concept by sfan5 08:57 ShadowBot https://github.com/luanti-org/luanti/issues/17602 -- Use texture pack name from conf when present by ZenonSeth 08:57 ShadowBot https://github.com/luanti-org/luanti/issues/17595 -- Fix light for liquid source faces below solid nodes by Desour 08:58 sfan5 uhh I'm not merging #17615 of course. meant #17582 08:58 ShadowBot https://github.com/luanti-org/luanti/issues/17615 -- [don't squash] Some refactoring + Lua CLI proof of concept by sfan5 08:58 ShadowBot https://github.com/luanti-org/luanti/issues/17582 -- [no sq] Update Android stuff and other fixes by sfan5 09:27 [MatrxMT] Ping sfan5 09:29 sfan5 pong 19:37 sfan5 it tried to implement a new collision feature and got stuck with a hard to solve backwards-compatibility problem. also my code doesn't even work. 19:37 sfan5 s/^it/I/ 19:38 Krock welcome to physics 19:38 MTDiscord oh hi Krock 19:38 Krock been there, was hurt. 19:38 Krock hello @luatic 19:38 MTDiscord i'm thinking we should try to move to strictly linear interpolation with smaller dtimes 19:39 MTDiscord new collision feature? 19:39 MTDiscord this will require optimizing the physics step, but it should be highly parallelizable 19:40 Krock The current physics are already quite good. Executing it on "best possible" dtime intervals seems to be the largest challenge 19:41 MTDiscord commented on #17612 re possibility of too small dtime steps 19:41 ShadowBot https://github.com/luanti-org/luanti/issues/17612 -- Collision: Run parabola movement in two iterations by SmallJoker 19:42 MTDiscord Krock: mhm i don't think the current physics are good in comparison to what dedicated physics engines offer 19:42 MTDiscord quite honestly we don't have much physics except for SUVAT and AABB collisions at all 19:43 MTDiscord proper physics means more colliders for example 19:43 MTDiscord a concept of mass, of forces, of energy and of collisions actually resulting in changes to velocity beyond "an axis gets zero'd" 19:43 MTDiscord also rotational velocity, for example 19:43 Krock the client code did steps of 0.1 * BS or so, which introduced such precision issues 19:44 Krock ah still. > f32 position_max_increment = 0.1*BS; 19:44 MTDiscord Krock: my worry is that if you have a peak very close to the start or end of the interval, you can get an arbitrarily small "epsilon-step" 19:45 MTDiscord <.zenonseth> there was an issue a while back about this, iirc, and i left comments - but i personally think adding a sphere collider, without adding mass, forces, energy, etc is doable, and can result in sliding/slip by on abitrary axis of collision somewhat easily 19:45 MTDiscord say the parabola peaks at dt=1e-7, the proposed change will issue a collisionMoveSimple with that dt then which might be a problem 19:45 Krock extreme cases already filtered out with the stupid simple (acceleration * speed < -0.1f), where I could raise the bar a bit 19:45 Krock it might be a problem for 1% of the simulation steps, which I do honestly not consider a problem 19:46 MTDiscord Zenon Seth: absolutely, these issues can be solved somewhat independently, and it makes sense to approach it like that. 19:46 MTDiscord though personally i don't care as much about spheres as i care about boxes and triangles ;) 19:47 sfan5 does anyone have any thoughts on introducing collision groups so that objects can choose to collide with? 19:48 MTDiscord it's a good feature, didn't we have an issue for that? 19:48 sfan5 I wanted to make a set_properties({collide_with_players=false}) work but it basically boils down to this 19:48 sfan5 dunno 19:48 MTDiscord i'm pretty sure someone linked how godot does it in some issue 19:49 sfan5 https://github.com/godotengine/godot/issues/7793 I suppose 19:50 sfan5 ...which is pretty much exactly what chatgpt suggested ... and I then tried to implement but got some detail wrong 19:51 MTDiscord <.zenonseth> er tangentially related btw but a while back (so its out of date) I wasted like 4 hours with claude making summaries of all issues: - Summaries: https://contentdb.milcho.org/misc/all_summaries.md - And high confidence duplicates: https://contentdb.milcho.org/misc/verified_high_confidence.md 19:59 k327- my server (luplay.k327.eu:30000) has had issues with being temporarily online on the server list for a while, getting a 409 error from servers.luanti.org/announce in the logs, but today that's been happening non-stop since 7 hours and still going on. I asked about it on the matrix room and got redirected by wsor to ask sfan5 here on irc 20:00 k327- offline* 20:00 sfan5 well, did you read the error messagE? 20:01 sfan5 because it should say what the problem is 20:02 k327- HTTPFetch for https://servers.luanti.org/announce returned response code 409 20:02 k327- Response body: 20:02 k327- Request has been filed, but the previous one encountered an error: 20:02 k327- Server luplay.k327.eu port 30000 did not respond to ping (tried 212.127.78.214) 20:03 Krock @.zenonseth It seems the AI is also a little bit drunk in some places "- #6527 - Move interact code from game.cpp to an own file" 20:03 ShadowBot https://github.com/luanti-org/luanti/issues/6527 -- Formspec replacement 20:04 sfan5 k327-: yes. did you ensure that your server is reachable on this IP and port? 20:05 k327- yes, you can even try joining it yourself 20:06 MTDiscord <.zenonseth> Krock: possibly, i ran it with the intention of finding duplicates, I have a much much longer jsonl (~2mb) with all summaries of first posy budy which could also be used for finding duplicates or seraching 20:06 MTDiscord lol, it doesnt work 20:07 MTDiscord my bad for assuming you actually checked