Time Nick Message 01:16 [MatrxMT] an LLM would make up the values to put into the tables even if it didn't know them. Several times I've seen people say "AI told me this API/field exists" and I've said "nope" 01:55 user333_ Blockhead: yeah, AIs lie a lot, back in like ~2021 it told me to directly wire a microcontroller to 120VAC and use a non-existant getBusVoltage() function to test for power... great idea! 02:51 repetitivestrain sfan5: thanks! 02:51 repetitivestrain (regarding #16634) 02:51 ShadowBot https://github.com/luanti-org/luanti/issues/16634 -- Automatically choose multiple emerge threads for singlenode by sfan5 03:16 repetitivestrain !tell lofhansl: regarding https://github.com/luanti-org/luanti/issues/15219: the client and server will be at variance anyway if the entity's physics are any more sophisticated than integrating velocity and engine-provided acceleration with respect to dtime, as the granularity afforded by their respective event loops will be inconsistent, the client cannot reproduce whatever physics routines exist on the server, and scarce few mods 03:16 MinetestBot repetitivestrain: Do not put an : at the end of nickname 03:16 repetitivestrain implement physics that maintain consistency of the adjustments they apply in the face of variable dtimes. for example, https://github.com/sorcerykid/mobs/blob/master/init.lua#L1276 does not account for dtimes at all, and others fail dismally to account for the most elementary derivative rule, x^n = nx: https://github.com/mt-mods/deltaglider/blob/master/init.lua#L529. c.f. 03:16 repetitivestrain https://codeberg.org/mineclonia/mineclonia/src/branch/main/mods/PLAYER/playerphysics/elytra.lua#L206, https://codeberg.org/mineclonia/mineclonia/src/branch/main/mods/ENTITIES/mcl_mobs/physics.lua#L1318. what is problematic is when globalstep intervals are so protracted that it is no longer possible for an entity's correctly implemented physics routines to adjust its velocity in time to inhibit the engine from moving it an excessive 03:16 repetitivestrain distance _on the server_ (e.g., by applying gravity after a jump, failing which mobs are liable to jump over fences), which i trust you'll agree is more catastrophic than temporary desynchronization between clients and the server, and this may be satisfactorily addressed by an upper bound on the dtime 03:17 repetitivestrain !tell lhofhansl regarding https://github.com/luanti-org/luanti/issues/15219: the client and server will be at variance anyway if the entity's physics are any more sophisticated than integrating velocity and engine-provided acceleration with respect to dtime, as the granularity afforded by their respective event loops will be inconsistent, the client cannot reproduce whatever physics routines exist on the server, and scarce few mods 03:17 MinetestBot repetitivestrain: I'll pass that on when lhofhansl is around 03:17 repetitivestrain implement physics that maintain consistency of the adjustments they apply in the face of variable dtimes. for example, https://github.com/sorcerykid/mobs/blob/master/init.lua#L1276 does not account for dtimes at all, and others fail dismally to account for the most elementary derivative rule, x^n = nx: https://github.com/mt-mods/deltaglider/blob/master/init.lua#L529. c.f. 03:17 repetitivestrain https://codeberg.org/mineclonia/mineclonia/src/branch/main/mods/PLAYER/playerphysics/elytra.lua#L206, https://codeberg.org/mineclonia/mineclonia/src/branch/main/mods/ENTITIES/mcl_mobs/physics.lua#L1318. what is problematic is when globalstep intervals are so protracted that it is no longer possible for an entity's correctly implemented physics routines to adjust its velocity in time to inhibit the engine from moving it an excessive 03:17 repetitivestrain distance _on the server_ (e.g., by applying gravity after a jump, failing which mobs are liable to jump over fences), which i trust you'll agree is more catastrophic than temporary desynchronization between clients and the server, and this may be satisfactorily addressed by an upper bound on the dtime 09:40 sfan5 repetitivestrain: so https://github.com/luanti-org/luanti/issues/15219? 09:41 repetitivestrain sfan5: yes, i was responding to lhofhansl's last comment in the issue 09:44 sfan5 oh 09:44 sfan5 he's more likely to see it if you can convince someone to copy-paste your text there 09:45 repetitivestrain hmm, i'll try to persuade the person who's been reporting physics anomalies in mineclonia with high dtimes on the forums 12:10 MinetestBot 02[git] 04cx384 -> 03luanti-org/luanti: Remove unused EventReceiver 134cbe7b2 https://github.com/luanti-org/luanti/commit/4cbe7b2597b553bac5c457bdc8179b020557a6c4 (152025-11-14T12:09:39Z) 12:10 MinetestBot 02[git] 04sfan5 -> 03luanti-org/luanti: More minor shader cleanups 1300eea43 https://github.com/luanti-org/luanti/commit/00eea43131bec7afc8321b88e4fc77e80594ce8e (152025-11-14T12:10:24Z) 12:10 MinetestBot 02[git] 04sfan5 -> 03luanti-org/luanti: Adjust shader test CI workflows 13577fc44 https://github.com/luanti-org/luanti/commit/577fc44ea9fd3c0d2ae657c53cddeb06bb05abd6 (152025-11-14T12:10:24Z) 15:29 cheapie I should dig the HP t520 back out, see what Luanti runs like on there now that the array texture stuff all works, seeing as how that thing does do OpenGL 4.6 15:29 cheapie I would say t420 instead but that only has 2GiB of RAM (soldered), while in the t520 I can add enough RAM for VE-C 15:30 [MatrxMT] the t520 in question: There's an OpenGL 4 now!!? 15:30 cheapie It supports Vulkan 1.2 too, at least in Linux... actually both do 17:45 [MatrxMT] i think the maximum size for nametags is too big 17:45 [MatrxMT] it gets too much in your face 17:45 MinetestBot lhofhansl: Nov-14 03:17 UTC regarding https://github.com/luanti-org/luanti/issues/15219: the client and server will be at variance anyway if the entity's physics are any more sophisticated than integrating velocity and engine-provided acceleration with respect to dtime, as the granularity afforded by their respective event loops will be inconsistent, the client cannot reproduce whatever physics routines exist on the ser 17:54 MinetestBot 02[git] 04lhofhansl -> 03luanti-org/luanti: Force highp precision for float and sampler2d on GLES, if supported (… 13785042d https://github.com/luanti-org/luanti/commit/785042dc9ebde9a61e6989425ee4e7026833ea1f (152025-11-14T17:52:55Z) 18:15 MTDiscord lua_api.md, line 4389 there's an example translated chatcommand "playtime". In the return it accesses a variable "minutes" instead of "playtime". Currently line 4396: minutes, tostring(minutes)) Should be playtime, tostring(playtime)) 18:15 MTDiscord (Or I don't understand translations. 50:50 chance) 19:33 [MatrxMT] It's a typo 19:34 [MatrxMT] (Although I do recall testing said code when writing the documentation; idk what went wrong in the process) 19:55 cheapie new billboard works great: https://cheapiesystems.com/media/images/screenshot_20251114_135500.png 19:56 cheapie other side: https://cheapiesystems.com/media/images/screenshot_20251114_134713.png 19:58 MTDiscord amazing 20:00 cheapie Some other player built that, but didn't actually put anything on it, so it was just a blank screen... so I modified my "image2table" script to automatically split across digiscreen nodes and to output the boilerplate stuff for me, then I could just copy/paste a column at a time into a Luacontroller until all 11 columns were lit up 20:01 MTDiscord It's some base64 images framework? 20:01 cheapie It's a whole bunch of digiscreens, they take images as nested tables of hex color values 20:01 MTDiscord Oh 20:02 cheapie The digistuff GPU can accept base64 strings (cheaper to send) and convert them 20:02 [MatrxMT] this seems like a usecase for a "texture screen" node 20:02 [MatrxMT] so you can input a base64 image into it 20:03 cheapie That's largely what these end up doing, those nested tables get turned into a base64-ed PNG for display 20:04 cheapie Initially with compression level 1, then if the image doesn't change for 5 seconds they go back and recompress at level 9 (and throw out the uncompressed copy), async both times 22:40 cheapie This should be enough pixels for now: https://cheapiesystems.com/media/images/screenshot_20251114_163731.png