| Time |
Nick |
Message |
| 00:13 |
|
Verticen joined #luanti |
| 00:23 |
|
FileX joined #luanti |
| 00:27 |
|
fluxionary joined #luanti |
| 00:42 |
|
est31 joined #luanti |
| 00:45 |
|
qqe joined #luanti |
| 01:11 |
|
user2_ joined #luanti |
| 01:26 |
|
FileX joined #luanti |
| 01:29 |
|
user2_ joined #luanti |
| 01:34 |
|
user2_ joined #luanti |
| 01:42 |
|
user2_ joined #luanti |
| 01:44 |
|
est joined #luanti |
| 02:33 |
|
Kimapr joined #luanti |
| 02:43 |
[MatrxMT] |
<Blockhead256> https://forum.luanti.org/viewtopic.php?t=23528 |
| 02:43 |
[MatrxMT] |
<Blockhead256> rather https://content.luanti.org/packages/Emojiminetest/linetrack/ |
| 02:44 |
|
Verticen joined #luanti |
| 02:44 |
[MatrxMT] |
<Blockhead256> the system works just like Advtrains tracks |
| 02:45 |
|
aliasstilltaken joined #luanti |
| 02:54 |
repetitivestrain |
amusing bug: it is now possible to edit the version text in the pause menu |
| 02:54 |
[MatrxMT] |
<Blockhead256> formspec prepend? I think it is known that that can override the pause menu |
| 02:55 |
repetitivestrain |
no, it's actually become an editable text input widget |
| 02:55 |
[MatrxMT] |
<Blockhead256> lol |
| 03:08 |
|
FileX joined #luanti |
| 03:34 |
cheapie |
For me it has a cursor and I can move it around but not edit anything with it |
| 03:35 |
cheapie |
Shame I can't edit it into 5.15 :P |
| 03:36 |
[MatrxMT] |
<Blockhead256> 5.15? you're as far ahead as the people with xbox 720 or 1080 in 2006 |
| 04:00 |
|
MTDiscord joined #luanti |
| 04:29 |
|
FeXoR joined #luanti |
| 04:36 |
|
Kimapr_ joined #luanti |
| 04:51 |
|
SFENCE joined #luanti |
| 05:20 |
|
SFENCE joined #luanti |
| 05:33 |
repetitivestrain |
there is a cursor and editing works, but not delete or shift |
| 05:33 |
repetitivestrain |
or backspace |
| 05:47 |
|
kaeza joined #luanti |
| 05:49 |
|
identity joined #luanti |
| 05:58 |
|
FeXoR joined #luanti |
| 05:58 |
|
SFENCE joined #luanti |
| 06:01 |
|
SFENCE joined #luanti |
| 06:48 |
|
YuGiOhJCJ joined #luanti |
| 07:41 |
|
jluc joined #luanti |
| 08:01 |
|
jaca122 joined #luanti |
| 08:17 |
|
Kimapr joined #luanti |
| 08:22 |
|
Warr1024 joined #luanti |
| 08:47 |
|
Warr1024 joined #luanti |
| 09:11 |
|
Kimapr_ joined #luanti |
| 09:14 |
|
FileX joined #luanti |
| 09:26 |
|
Kimapr joined #luanti |
| 09:31 |
|
Verticen joined #luanti |
| 09:31 |
|
FileX joined #luanti |
| 09:41 |
|
Kimapr joined #luanti |
| 09:44 |
|
Talkless joined #luanti |
| 10:00 |
|
FileX joined #luanti |
| 10:27 |
|
fluxionary joined #luanti |
| 10:53 |
|
Haven0320 joined #luanti |
| 11:17 |
|
Verticen joined #luanti |
| 11:34 |
|
FileX joined #luanti |
| 12:17 |
|
identity joined #luanti |
| 12:22 |
|
PoochInquisitor joined #luanti |
| 12:46 |
|
Kimapr_ joined #luanti |
| 13:04 |
|
SFENCE joined #luanti |
| 13:15 |
|
CRISPR joined #luanti |
| 13:19 |
|
user2_ joined #luanti |
| 13:23 |
|
erle joined #luanti |
| 13:26 |
|
kimapr__ joined #luanti |
| 13:35 |
|
Menchers joined #luanti |
| 13:46 |
|
kimapr__ joined #luanti |
| 14:04 |
user2_ |
yeah, i just confirmed you can mess with the text in the pause menu- you can add characters but not delete them |
| 14:06 |
user2_ |
it was probably caused by commit 9972639 as it didnt work before that- |
| 14:07 |
user2_ |
that is a very interesting bug- |
| 14:16 |
|
MTDiscord1 joined #luanti |
| 14:17 |
|
Helenah joined #luanti |
| 14:27 |
|
fluxionary joined #luanti |
| 14:35 |
[MatrxMT] |
<y5nw> cheapie: You can edit it into 5.15 in the overwrite mode (press the insert key) |
| 14:36 |
[MatrxMT] |
<y5nw> Looks like I did not test the PR thoroughly enough when reviewing it; oops. |
| 14:37 |
user2_ |
i now have "Lu-non-ti 5.15.1-dev" in my pause menu ;-; |
| 14:44 |
|
MTDiscord2 joined #luanti |
| 14:47 |
[MatrxMT] |
<y5nw> #16504 |
| 14:47 |
ShadowBot |
https://github.com/luanti-org/luanti/issues/16504 -- Add missing IsWritable checks by y5nw |
| 14:55 |
|
Kimapr joined #luanti |
| 14:56 |
|
CRISPR joined #luanti |
| 14:58 |
|
Kimapr_ joined #luanti |
| 15:02 |
user2_ |
seems to have fixed it |
| 15:02 |
|
Verticen joined #luanti |
| 15:09 |
|
Kimapr joined #luanti |
| 15:14 |
|
Kimapr joined #luanti |
| 15:19 |
|
mrcheese joined #luanti |
| 15:40 |
|
identity joined #luanti |
| 16:11 |
|
Kimapr joined #luanti |
| 16:35 |
cheapie |
y5nw: I see https://cheapiesystems.com/media/images/screenshot_20250921_113502.png |
| 16:36 |
cheapie |
Interestingly shift seems to be ignored, the characters come out lowercase anyway |
| 16:37 |
|
mrcheese_ joined #luanti |
| 16:45 |
|
CRISPR joined #luanti |
| 16:45 |
|
mrcheese joined #luanti |
| 16:48 |
[MatrxMT] |
<birdlover32767> i am trying to clone with git clone and have been unsucsessful/cloned the wrong thing several times now |
| 16:49 |
[MatrxMT] |
<birdlover32767> github is a hassle |
| 16:49 |
[MatrxMT] |
<Blockhead256> well git has got a sharp learning curve for sure |
| 16:50 |
user2_ |
how so? i just use 'git clone --depth 1 -b master https://github.com/luanti-org/luanti.git' |
| 16:50 |
cheapie |
y5nw: OK, I've had my fun, you can fix it now: https://cheapiesystems.com/media/images/screenshot_20250921_114752.png |
| 16:50 |
user2_ |
i also use git to manage my backups- its great :D |
| 16:50 |
[MatrxMT] |
<Blockhead256> user2_: why are you specifying a branch though? The default branch doesn't need to be named by the end user, that's what git will fetch and checkout by default |
| 16:51 |
user2_ |
just by habit, i tend to do -b stable-5 often |
| 16:52 |
[MatrxMT] |
<Blockhead256> birdlover32767: all you need to do is copy and paste the URL to the main page of any github repo. So: https://github.com/luanti-org/luanti becomes `git clone https://github.com/luanti-org/luanti` |
| 16:54 |
user2_ |
it doesnt have to be a github repo in specific- it can also be gitlab/another repo folder on your PC |
| 16:54 |
user2_ |
(and most of the other git platforms) |
| 16:54 |
[MatrxMT] |
<Blockhead256> correct, this explanation just points out one very specific action. Same idea on GitLab, gitea, ... |
| 16:55 |
user2_ |
yeah, for the most basic users, 'git clone <github repo URL>' should be enough |
| 16:56 |
|
mrcheese joined #luanti |
| 16:58 |
[MatrxMT] |
<birdlover32767> first my internet went out while cloning, then i accidentally copied from the link in the readme.md, then i accidentally cloned from master instead of some other branch |
| 16:59 |
[MatrxMT] |
<birdlover32767> so it's more my fault tbh |
| 16:59 |
[MatrxMT] |
<Blockhead256> well, it's easier to check out other branches than to re-clone the whole thing |
| 16:59 |
[MatrxMT] |
<Blockhead256> `git checkout -b <branchname>` |
| 17:00 |
[MatrxMT] |
<birdlover32767> ...i was using the github webpage |
| 17:00 |
[MatrxMT] |
<Blockhead256> downloading zip files? |
| 17:00 |
[MatrxMT] |
<birdlover32767> cloning .git files |
| 17:00 |
[MatrxMT] |
<Blockhead256> well on the bright side, you learned several things not to do next time |
| 17:01 |
[MatrxMT] |
<y5nw> cheapie: I suppose you don't mind being featured on #16504? |
| 17:01 |
ShadowBot |
https://github.com/luanti-org/luanti/issues/16504 -- Add missing IsWritable checks by y5nw |
| 17:01 |
user2_ |
:O i just got 5.1.1 to compile on linux |
| 17:01 |
[MatrxMT] |
<birdlover32767> why 5.1.1 |
| 17:01 |
cheapie |
y5nw: Go for it :P |
| 17:01 |
user2_ |
it's very broken- and i want to see if i can get all 5.x versions on linux |
| 17:02 |
[MatrxMT] |
<y5nw> cheapie: Already done (-> "Gallery") |
| 17:02 |
* cheapie |
nods, saw it |
| 17:02 |
user2_ |
had to modify a file and specify like 5 options for cmake- and watch 5000 warnings scroll by |
| 17:03 |
[MatrxMT] |
<birdlover32767> also for some reason it did not clone some submodules even though i specified --recursive |
| 17:03 |
user2_ |
uh- iirc you dont need to specify --recursive... |
| 17:03 |
[MatrxMT] |
<birdlover32767> so now i have to reinstall them... ugh |
| 17:03 |
[MatrxMT] |
<birdlover32767> reclone* |
| 17:04 |
user2_ |
only some repos need --recursive- not luanti-org/luanti tho |
| 17:04 |
[MatrxMT] |
<y5nw> cheapie: Done |
| 17:06 |
user2_ |
you can also use wget to download a specific release- 'wget https://github.com/luanti-org/luanti/archive/refs/tags/<version>.zip' |
| 17:12 |
|
sparky4 joined #luanti |
| 17:13 |
|
behale_ joined #luanti |
| 17:14 |
|
diceLibr- joined #luanti |
| 17:16 |
|
identity joined #luanti |
| 17:20 |
user2_ |
got 5.1.0 to compile as well- |
| 17:21 |
Krock |
if you need old binaries - get the Windows' builds and run them in Wine if you don't have Windows |
| 17:21 |
[MatrxMT] |
<Blockhead256> win32 is the most stable API on Linux :rollsafe: |
| 17:22 |
Krock |
ironically those are the most compatible in terms of ABI |
| 17:22 |
user2_ |
i want to run them properly- also wine doesn't work for me on <=5.5.0 |
| 17:22 |
user2_ |
also, using wine is a lot less fun then compiling :D |
| 17:24 |
|
Menchers joined #luanti |
| 17:28 |
|
kaeza joined #luanti |
| 17:28 |
user2_ |
well 5.1.0 works- i can play a 5.14.0-dev server with it |
| 17:30 |
user2_ |
trying 5.0.1 now- |
| 17:31 |
user2_ |
soo many warnings- |
| 17:39 |
user2_ |
5.0.1 compiled- i'm almost in 0.4x territory ;-; |
| 17:44 |
erle |
user2_ i am curious. 1 do you intend to document your journey? 2. what is your solution to the “we have submodules at home” situation with irrlichtmt? |
| 17:46 |
user2_ |
it's pretty easy tbh- download the zip of the version's src code, unzip it, add '#include <memory>' to src/clientiface.h, and use a couple options with CMake, hold on... |
| 17:47 |
user2_ |
here are the options for CMake: -DRUN_IN_PLACE=TRUE -DCMAKE_CXX_FLAGS="-std=c++03" -DENABLE_LUAJIT=OFF -DENABLE_SOUND=OFF |
| 17:47 |
user2_ |
i turned off luajit and sound to avoid any issues with them |
| 17:48 |
user2_ |
for 5.4.1-5.2.0 you only need to add '#include <memory>' in clientiface.h |
| 17:50 |
user2_ |
for dependencies, they actually weren't an issue, just copy/paste the list from README.md and it works |
| 17:55 |
erle |
user2_ you might want to disable LTO in later builds unless you want to wait forever for little benefit |
| 17:56 |
user2_ |
linking takes like 3 seconds for them right now |
| 17:56 |
user2_ |
so i don't think it will be an issue |
| 17:57 |
|
Desour joined #luanti |
| 18:02 |
|
SFENCE joined #luanti |
| 18:02 |
erle |
user2_ older builds do not have LTO enabled by default. but with LTO, linking takes ages and consumes a lot more memory. depends on your machine if this is a problem though. while i own no machine on which it would not be an issue, some people apparently don't notice such overhead. |
| 18:04 |
user2_ |
i have 8GB of RAM and a 4-core CPU @2.4 GHZ- it compiles in ~10min with all 4 cores |
| 18:06 |
cheapie |
I have 64GiB and a 16-core CPU @ 4ishGHz, current version compiles client+server in 57 seconds |
| 18:06 |
user2_ |
my old PC has 2GB of RAM and a 2-core CPU @1.4 GHZ- takes ~30min to compile and gets ~0.3 FPS ;-; |
| 18:07 |
user2_ |
5.0.0 just compiled, ima see if it runs |
| 18:08 |
luatic |
how big of a problem LTO is probably depends a lot on your linker, with mold it might be acceptable. but i usually disable it as well. |
| 18:09 |
user2_ |
compiling doesn't take too long for me so idc too much |
| 18:10 |
user2_ |
one thing i have noticed is newer clients (especially 5.11+) get much higher FPS than older ones |
| 18:11 |
erle |
user2_ 0.3 fps, what graphics are you running with it? |
| 18:11 |
user2_ |
uhh- the internal GPU of an ancient laptop (intel graphics) |
| 18:11 |
erle |
user2_ what one? |
| 18:11 |
user2_ |
i'm not sure tbh |
| 18:12 |
user2_ |
5.1.1 got ~20FPS in windows XP, but 5.13.0 got 0.3 FPS- |
| 18:12 |
erle |
oh, that's easy |
| 18:12 |
erle |
they removed the opengl 1.x renderer |
| 18:12 |
user2_ |
yeah, that's the issue- |
| 18:13 |
user2_ |
that was removed in... 5.10 iirc? |
| 18:13 |
erle |
i think 5.11? |
| 18:13 |
erle |
this means you can't play it on old hardware anymore and on newer (but weak) hardware that nominally supports newer opengl still gets a noticeable frame drop |
| 18:13 |
user2_ |
yeah, this has opengl 1.4.0 ;-; |
| 18:13 |
user2_ |
my main PC supports opengl 3.1 |
| 18:14 |
erle |
the issue is not about what it supports |
| 18:14 |
erle |
i have several computers that support newer opengl, but older opengl runs better performance-wise |
| 18:14 |
erle |
it is not productive to just look at what the specs support |
| 18:14 |
user2_ |
older opengl tends to have more visual bugs from my experience |
| 18:15 |
erle |
wdym? |
| 18:15 |
|
SFENCE joined #luanti |
| 18:15 |
user2_ |
i tend to get more visual bugs on older opengl... |
| 18:15 |
erle |
for example, several intel integrated chipsets actually can support opengl 2.0 but mesa artificially limits it to 1.4 or 1.5 because otherwise chrome takes a rendering part that is less efficient on that hardware (it defaults to newer opengl) |
| 18:15 |
erle |
s/part/path/ |
| 18:16 |
user2_ |
i'm pretty sure this can't use newer opengl- as in no hardware support |
| 18:16 |
user2_ |
luantiserver runs perfectly tho |
| 18:17 |
user2_ |
well, i got all 5.x versions running on linux, so that goal is achieved |
| 18:17 |
* SwissalpS |
claps |
| 18:18 |
|
Kimapr joined #luanti |
| 18:18 |
user2_ |
(except for the 5.8.1 android version ofc, that's just a bugfix for android) |
| 18:19 |
erle |
luatic i think the irony is that if you use LTO for games that you run on the same machine (i.e. not a release build), most machines that have no problems doing the LTO (i.e. have enough RAM and CPU to spare) won't benefit much from the savings. |
| 18:19 |
erle |
luatic meanwhile, machines that could benefit in some way from the LTO don't have the resources to do it themselves |
| 18:19 |
luatic |
well yes, LTO makes most sense for release builds intended for distribution |
| 18:20 |
erle |
you can actually see this with some other game engine stuff, where building the thing has higher requirements than running it. if LTO is enabled |
| 18:20 |
erle |
i think godot has LTO enabled by default too |
| 18:20 |
erle |
bizarre decision, but again “gamers who are devs have the best possible hardware” |
| 18:20 |
user2_ |
i get modern versions to build/link perfectly on my RPi5 |
| 18:27 |
erle |
user2_ btw, enable mipmaps. they look shit (because of a long-standing unwillingness to scale graphics in a gamma-correct way), but they can improve performance IIRC. |
| 18:27 |
|
Thermoriax joined #luanti |
| 18:29 |
Desour |
erle: can you actually measure a performance difference with mipmapping for textures as small as in luanti usually? |
| 18:30 |
erle |
Desour depends on view range obviously. though i, as always, ask people to perform their own measurements. and not with mesa software renderer. the software renderer knows no limits excepd the thermal throttling on your CPU. |
| 18:32 |
erle |
in fact, i think if a lot more people would actually try to test their graphics code instead of conjecturing, there would be a lot fewer rendering problems in a lot of software. |
| 18:32 |
Desour |
erle: yeah, but I mean on one of your machines with gpu opengl |
| 18:33 |
erle |
i think it's the same as the triangle merging thing: if your machine is “good enough”, you won't notice it |
| 18:33 |
erle |
give me a moment |
| 18:37 |
Desour |
the triangle merging thing (it was called "fast faces") caused fewer triangles, but same count of mesh buffers. so yes, it is possible that it makes things slower on machines where luanti is gpu bound |
| 18:43 |
erle |
i think it was not removed because of that though, but because no one could find it making a tangible difference on coredev hardware |
| 18:44 |
erle |
also unless i am remembering it wrong one triangle in a strip always was different (had the wrong winding or something), which may have inhibited a trivial lighting system improvement |
| 18:44 |
erle |
although i never saw anyone except lizzy and me discussing this |
| 18:44 |
sfan5 |
because it's a waste of time to discuss |
| 18:45 |
Desour |
removing it makes things slower* I meant |
| 18:46 |
erle |
i think you misunderstand. now that you don't have the triangle merging, that solution could be used … if you still used irrlicht. |
| 18:46 |
erle |
in any case, i'll check the mipmap performance now |
| 18:49 |
sfan5 |
we are not restricted in any way for performance improvements just because we're no longer using irrlicht (upstream) |
| 18:49 |
sfan5 |
I don't understand this argument |
| 18:50 |
user2_ |
does 'moving away from irrlicht' mean 'no more plain GUIs?' |
| 18:50 |
user2_ |
if so, :D |
| 18:52 |
erle |
sfan5 if you were using irrlicht, you could use irrlicht-specific code without changing it, that's all. |
| 18:55 |
erle |
Desour with 5.10 and mineclonia (idk what version), looking at a complex landscape i increased view range until i got like 20 fps. i enabled mipmaps and rejoined that exact scene, i got 22fps. i verified it has no effect when the shown stuff is all close, i.e. in piranesi, as few rooms are big enough mipmaps are used. |
| 18:56 |
erle |
idk if that can be generalized, but 10% fps more ore less is not nothing, given that if these mipmaps were actually scaled correctly there would be little loss of fidelity |
| 18:57 |
erle |
in any case, as almost all performance things, you only notice them when you are right at the edge of playable/stuttering |
| 19:01 |
Desour |
erle: did you also try rejoining with mipmap disabled, to rule out more mapblocks being loaded the first time due to rotating the camera or slight movement? |
| 19:02 |
Desour |
but interesting that there apparently is a small difference for you :eyes: |
| 19:02 |
erle |
Desour i did actually move the first and second times and waited until the fps settled |
| 19:02 |
erle |
Desour what is interesting about it? was it unexpected somehow? |
| 19:04 |
Desour |
it's interesting to have this datapoint |
| 19:05 |
Desour |
I didn't expect one way or the other. on new machines I would be surprised though |
| 19:05 |
erle |
this is a machine from 2018. |
| 19:05 |
erle |
oh wait no, hmm |
| 19:05 |
erle |
from 2021 actually |
| 19:05 |
erle |
new enough? |
| 19:06 |
Desour |
that's newer than my machine |
| 19:06 |
user2_ |
my RPi5 is from 2023, but has specs from 2010 |
| 19:06 |
Desour |
erle: so what gpu and cpu did you use? |
| 19:06 |
user2_ |
more like ~2014 |
| 19:06 |
erle |
Desour imx8mq with vivante (7000?) GPU |
| 19:07 |
erle |
it has like 4 aarch64 CPUs |
| 19:08 |
user2_ |
muy RPi5 has a 4-core ARM Cortex-A76 @2.4 GHZ and 8GB of LP-DDR4 RAM |
| 19:08 |
erle |
ironically, the GPU driver is a bit broken right now (i am running debian unstable on it and i finally know what “unstable” means) |
| 19:08 |
user2_ |
*my |
| 19:08 |
erle |
but it only affects non-minetest |
| 19:10 |
sfan5 |
comparing with "normal" x86 machines that's probably more like 2012-level of performance |
| 19:10 |
sfan5 |
minus a few special usecases the vendor cared about |
| 19:12 |
erle |
well, it's one of the things that can do opengl es 3.x (allegedly), but desktop opengl 1.x is noticeably faster for some workloads (e.g. minetest/luanti) |
| 19:13 |
Desour |
erle: what does glxinfo -B (and/or eglinfo and vulkaninfo) say? |
| 19:13 |
|
lhofhansl joined #luanti |
| 19:13 |
erle |
but in fact, i can run circles around it using a thinkpad T60 or R60 (from 2007) that has been upgraded with an SSD |
| 19:13 |
MinetestBot |
[git] y5nw -> luanti-org/luanti: Add missing IsWritable checks (#16504) 2eded3f https://github.com/luanti-org/luanti/commit/2eded3f078aa0924aeafe8607b3ac0d2e89f2dfb (2025-09-21T19:12:42Z) |
| 19:13 |
erle |
at least for minetest/luanti and until 5.10 (because with shaders the performance drops there as well) |
| 19:14 |
|
Kimapr joined #luanti |
| 19:15 |
erle |
Desour i'll check. why do you want to know? |
| 19:15 |
Desour |
erle: device name (you werent sure about the 7000). and maybe also supported gl and vulkan versions |
| 19:19 |
erle |
Desour it's vivante GC7000 rev 6214. it claims to do ogl 2.1 and ogles 2.0. so while nominally better than a 20 year old intel GPU, it doesn't actually outperform them for desktop opengl though. |
| 19:20 |
sfan5 |
https://www.nxp.com/products/processors-and-microcontrollers/arm-processors/i-mx-applications-processors/i-mx-8-applications-processors:IMX8-SERIES in case anyone was wondering what NXP thinks you should use these for |
| 19:20 |
sfan5 |
>Infotainment and In-Vehicle Experience |
| 19:20 |
erle |
the answer is probably “in-flight entertainment systems” |
| 19:20 |
erle |
hey, i was close! |
| 19:20 |
erle |
xD |
| 19:21 |
erle |
yeah, the bizarre choice of GPU means i *have* to use wayland |
| 19:21 |
luatic |
erle: my laptop is from about 2020, maybe 2019 i think, mid range, pretty average (for german standards) really i think, and runs much more decently |
| 19:21 |
erle |
because the vendor only supports that |
| 19:21 |
user2_ |
i found that <=5.4.1 REQUIRE x11 to run in client mode- |
| 19:21 |
luatic |
(i've got a ryzen 5 3500u) |
| 19:22 |
luatic |
well i have upgraded my RAM at some point, but that's mostly irrelevant for luanti, i did that for other reasons |
| 19:22 |
sfan5 |
luatic: you have to remember that erle doesn't have a conventional laptop but essentially embedded hardware (think: older RPIs) in a laptop shell |
| 19:22 |
user2_ |
i have a modern RPi :P |
| 19:23 |
erle |
i am pretty sure you *can* actually have the MNT reform with an rpi in it, one moment |
| 19:23 |
erle |
yeah lol https://shop.mntre.com/products/mnt-reform-cm4-processor-module-adapter |
| 19:25 |
erle |
i think when “conventional laptop” stopped meaning “any x86/x64 thinkpad that can run debian and has enough performance to watch youtube videos” i stopped having a conventional laptop ;) |
| 19:26 |
user2_ |
my 'laptop' is a potato x86 (not x86_64) that runs debian 12 |
| 19:26 |
user2_ |
it's technically i686 |
| 19:26 |
erle |
i think you have to have i686 nowadays otherwise a lot of stuff simply won't run |
| 19:27 |
erle |
i am writing you from, one moment |
| 19:27 |
user2_ |
pretty much every piece of software on this i had to compile myself |
| 19:27 |
erle |
Model name: Genuine Intel(R) CPU T2400 @ 1.83GHz |
| 19:27 |
|
SFENCE joined #luanti |
| 19:27 |
erle |
user2_ why? gentoo user? |
| 19:28 |
user2_ |
no, i use debian |
| 19:28 |
user2_ |
mostly because the APT packages were out of date/non-existant |
| 19:28 |
erle |
for debian 12? |
| 19:28 |
MinetestBot |
[git] Sokomine -> luanti-org/luanti: Add client-side logging of chat messages (#14289) afd681d https://github.com/luanti-org/luanti/commit/afd681d013720276aaa6307b8ff67d0afc6471b9 (2025-09-21T19:27:42Z) |
| 19:29 |
erle |
that should be still supported for a while now |
| 19:29 |
erle |
OH LOL |
| 19:29 |
erle |
FINALLY |
| 19:29 |
erle |
client-side logging of chat was one of the reasons i told people to use a non-standard client |
| 19:29 |
user2_ |
the package for minetest on debian 12 is minetest 5.6.1 ;-; |
| 19:30 |
luatic |
yeah forget about that |
| 19:30 |
user2_ |
debian 13 has 5.10 |
| 19:30 |
erle |
classic distribution thing |
| 19:31 |
user2_ |
and it's still called a 'multiplayer infinite-world sandbox game' why not a game engine >:-((( |
| 19:31 |
mrcheese |
lol still kinda a game |
| 19:31 |
user2_ |
not without at least MTG |
| 19:31 |
erle |
at least debian is faster than many other distributions. arch linux somehow successfully convinced its users that it has bleeding edge software, but in reality 2 out of 2 packages where i was upstream it took a LOT of time to get it right. |
| 19:32 |
user2_ |
debian 11 has... 5.3.0 or 5.4.0 iirc |
| 19:32 |
erle |
it has 5.3 |
| 19:32 |
mrcheese |
;-; |
| 19:33 |
erle |
in one case the arch package used a build script that was so wrong that i was able to alter the upstream package so that now the README on arch (but not another platform) simply is a re-telling about how the maintainer did not answer to my emails about description, dependencies, and build scripts being wrong for a year |
| 19:33 |
user2_ |
wasnt 5.3 the one with RCE exploits- |
| 19:33 |
erle |
tell me more |
| 19:33 |
mrcheese |
O_O |
| 19:33 |
erle |
it may have been the version where you could inject code into books |
| 19:34 |
mrcheese |
wow- |
| 19:34 |
erle |
the guilty people here in this channel know who they are |
| 19:34 |
mrcheese |
atleast its not the server executing code on the CLIENT |
| 19:34 |
erle |
client-sent SSM + server-sent CSM would be perfect balance |
| 19:34 |
erle |
just have both :P |
| 19:34 |
user2_ |
no, it was 5.1.1 |
| 19:35 |
erle |
maybe you are talking about a different RCE than i did |
| 19:35 |
erle |
explain? |
| 19:35 |
[MatrxMT] |
<y5nw> https://github.com/luanti-org/luanti/security/advisories/GHSA-7q63-4fq2-hqcr I suppose? |
| 19:35 |
|
SFENCE joined #luanti |
| 19:36 |
[MatrxMT] |
<y5nw> Given the implementation of core.deserialize and the serialization format it's not particularly surprising that it should not be used for user input |
| 19:36 |
MTDiscord |
<wsor4035> for lots of reasons, should just never use core.(de)serialize |
| 19:37 |
erle |
if only someone had suggested to everyone to read LANGSEC, especially the “seven turrets of babel” paper! |
| 19:37 |
|
SFENCE joined #luanti |
| 19:38 |
erle |
fun fact: unless i am mistaken, it is impossible to statically verify gltf json using json schema. make of that what you will :) |
| 19:39 |
luatic |
yeah no surprise if a file format is potent then regex / cfg doesn't suffice for validation..? |
| 19:39 |
luatic |
even at a type level |
| 19:40 |
luatic |
i mean quite obviously something like "the scene graph is free of cycles" can't be encoded in json schema |
| 19:40 |
luatic |
nevertheless i am very confident in tiniergltf + CGLTFMeshFileLoader |
| 19:40 |
luatic |
you should have seen tinygltf |
| 19:57 |
|
kamdard__ joined #luanti |
| 20:00 |
|
bwarden_ joined #luanti |
| 20:01 |
|
Sompi joined #luanti |
| 20:01 |
|
jonadab joined #luanti |
| 20:02 |
|
chilledfrogs joined #luanti |
| 20:03 |
|
peterz joined #luanti |
| 20:05 |
|
nekobit joined #luanti |
| 20:06 |
erle |
luatic no actually it is a surprise. |
| 20:07 |
erle |
luatic the things that i have seen that are not easy to validate are entirely due to the choice of data structures. |
| 20:09 |
erle |
luatic as an example: imagine you have an iot device list returned by an API, in JSON. assume each device has administrative data (e.g. who owns it, last maintenance date) and sensor data (e.g. temperature, humidity). |
| 20:10 |
erle |
luatic there exist two trivial ways to serialize this info. either as a list of devices that each have two sublists for admin and sensor data. or as a list of admin and sensor data that each contain entries for each device. the first thing can be easily validated using JSON schema. the second can not. |
| 20:12 |
|
SFENCE joined #luanti |
| 20:14 |
erle |
luatic i have an even simpler example that i often use to teach people how important it is to choose the right serialization. if you have a timespan, you can save it as two timestamps. however, then you have to calculate in your validator if the end is before the start. alternatively, you can use the start and and offset, which is WAY easier to validate (offset must be positive). |
| 20:21 |
MinetestBot |
[git] lhofhansl -> luanti-org/luanti: Monoblocks: optimize blocks that contain a single type of node (#16293) 08b7870 https://github.com/luanti-org/luanti/commit/08b7870c797909e934281679b6c064018fe9f07a (2025-09-21T20:19:30Z) |
| 20:21 |
|
mrcheese joined #luanti |
| 20:26 |
erle |
luatic the important insight here is of course that this can only go one way, you can only add computing power needs by designing a format badly. once you represent something that does not need context-sensitivity using something that is context-sensitive (e.g. an uncompressed bitmap vs RLE-encoded bitmap), you suddenly need a lot more computing power for validation, forever. |
| 20:27 |
erle |
RLE is not bad design though, i just chose it because it is obvious that it represents the same information as uncompressed bitmaps. |
| 20:27 |
erle |
while being context-sensitive, so a decoder suddenly needs more bounds checks etc. pp. |
| 20:29 |
user2_ |
time to update my client- |
| 20:29 |
erle |
‘quite obviously something like "the scene graph is free of cycles" can't be encoded in json schema” – it is quite possible to design data structures that do not allow for cycles by not allowing references, or only allowing them to go one way. |
| 20:29 |
erle |
(i.e. only go backwards, only go forwards, only go to a higher level, only access a definition …) |
| 20:33 |
user2_ |
turns out accidentally typing 'make -j44' instead of 'make -j4' makes your PC unhappy- |
| 20:36 |
MTDiscord |
<et086> why end all your messages in "-" |
| 20:36 |
user2_ |
why not- |
| 20:37 |
user2_ |
i could end them in... |
| 20:37 |
identity |
or in… |
| 20:38 |
user2_ |
or i could end them in\n |
| 20:43 |
mrcheese |
:P |
| 20:44 |
mrcheese |
or end them in_ |
| 20:46 |
user2_ |
or end them in the windows newline, which is iirc \r\n |
| 20:46 |
user2_ |
or \n\r |
| 20:46 |
user2_ |
idk |
| 20:47 |
SwissalpS |
better than ending all with 💩 :p |
| 20:48 |
user2_ |
that just renders as a square for me- i think it's my client |
| 20:49 |
identity |
that is a shit ending to be honest |
| 20:51 |
user2_ |
we should all use \n |
| 20:55 |
SwissalpS |
exactly identity :D |
| 20:55 |
user2_ |
windows and unix/linux hate each other enough to not use the same line endings |
| 21:00 |
MTDiscord |
<nathan4220776> I think you might be onto something. |
| 21:00 |
MTDiscord |
<nathan4220776> Not if you have enough cores. |
| 21:01 |
erle |
user2_ install gnu unifont |
| 21:02 |
erle |
user2_ then you can see shit emoji here: 💩 |
| 21:02 |
user2_ |
from where do i install it? |
| 21:02 |
user2_ |
is it an APT package? |
| 21:02 |
identity |
from your distro's package manager, something like “font-gnu-unifont” |
| 21:04 |
user2_ |
it's simply called 'unifont' |
| 21:05 |
erle |
user2_ afterwards, do “fc-cache -fv” |
| 21:05 |
erle |
and restart affected applications |
| 21:05 |
erle |
you should have simple 16 pixel high monochrome emojis in all applications then |
| 21:05 |
user2_ |
:D |
| 21:06 |
user2_ |
"fc-cache: succeeded" |
| 21:06 |
erle |
user2_ you have to install ttf-unifont though |
| 21:06 |
erle |
user2_ otherwise you only get unifont in .hex format, which is understood by the unifont toolchain and … my unicode_text mod :P |
| 21:07 |
user2_ |
i see it- is that a good thing? :P |
| 21:07 |
erle |
wdym? the .hex format is human-readable, but your applications will want TTF files i guess |
| 21:08 |
user2_ |
nvm |
| 21:08 |
user2_ |
i'l remove it if any problems arise |
| 21:08 |
identity |
you probably want a better font than unifont, though… |
| 21:09 |
Helenah |
user2_: https://ctrl-c.club/~helenah/story.html |
| 21:10 |
user2_ |
err- X |
| 21:10 |
user2_ |
D |
| 21:11 |
erle |
identity depends on what you mean with “better”. unifont is the only font i know that delivers clearly distinguishable monochrome emoji. it is also relatively small, a few MB. other icon fonts exist, but they are sometimes tens or hundreds of MB and do not actually have all characters. |
| 21:12 |
user2_ |
yeah- and we have ASCII emojis \_('_')_/ |
| 21:12 |
erle |
i think the only stuff that unifont is lacking is a) some egyptian hieroglyphs, like the “long penis” one b) emoji combinations/variants |
| 21:12 |
user2_ |
yeah we definitely need that one- |
| 21:12 |
identity |
unifont has “all the characters” but nobody said they were any good nor that they display correctly |
| 21:14 |
identity |
i can forgive the first one, but saying that your font has “full basic multilingual plane coverage” while most languages simply do not display anything close to correct is dishonest |
| 21:15 |
Helenah |
I don't think user2_ read the story... 8========D |
| 21:15 |
user2_ |
i don't really care what font i use- as long as basic text emojis work, i'm fine... |
| 21:15 |
Helenah |
Like the poop emoji? :3 |
| 21:16 |
Helenah |
It looks like a cow dung to me. |
| 21:24 |
|
CRISPR joined #luanti |
| 21:37 |
erle |
identity i think if the stuff appears close to correct is a question of the font engine really. like, unicode_text can not do arabic letter shaping, because i never found someone who a) was able to read/write arabic b) willing to assist me |
| 21:37 |
erle |
identity but yeah, i know about the difficulties. i have unifont set as default font for all my computers. |
| 21:38 |
|
frostsnow joined #luanti |
| 21:38 |
erle |
stuff like a transgender flag renders decomposed as its parts |
| 21:43 |
Helenah |
erle: Nancy Mace :3 |
| 21:47 |
erle |
Helenah what? |
| 21:48 |
Helenah |
erle: She's a mace against transgender women. |
| 21:48 |
Helenah |
She negs on about rape and transgender women. |
| 21:48 |
Helenah |
*nags |
| 21:48 |
Helenah |
I wish someone would stick a sock in it to be honest. |
| 21:48 |
erle |
Helenah well i think “look what that bigot said” is not exactly nice news ._. |
| 21:52 |
Helenah |
She keeps banging on about having been raped when she was younger, goes on about how it can take 25 years for a woman to come forward to say they been raped, because when they do, they're told they're bullshitting about it, and seems to group transgender women as men trying to take advantage of woman rights, and rape by predatory men, it seems to make me question "Have you actually been raped? |
| 21:52 |
Helenah |
Because you sure are obsessed with saying you are while claiming a 0.5% minority of the woman population are men and dangerous to women.", it really does make me question if it's true. |
| 21:52 |
Helenah |
erle: Yeah |
| 21:52 |
Helenah |
rofl |
| 21:53 |
erle |
Helenah the only good thing this mood-destroying paragraph did is remind me to take my estrogen |
| 21:53 |
Helenah |
If it is true, I think she needs help, shouldn't be that obsessed about it. |
| 21:54 |
Helenah |
erle: OH wait... |
| 21:54 |
Helenah |
How insensitive of me, you're transgender? SHIT, look... |
| 21:54 |
Helenah |
I'm sorry, I should've known better! |
| 21:55 |
erle |
apart from that, i just think “look what that asshole did online” is generally not something to dwell on |
| 21:55 |
erle |
or offline, so to say |
| 21:55 |
* mrcheese |
watches the conversation not be related to Luanti at all :PPP |
| 21:56 |
MTDiscord |
<nathan4220776> Unless, of course, you're looking for entertainment. |
| 21:56 |
Helenah |
How could this what I said possibly be entertaining? It's nerve tearing if anything. |
| 21:56 |
erle |
if your entertainment needs are met by “famous antivaxxer finally gets his shot”, i suggest https://old.reddit.com |
| 21:56 |
erle |
Helenah entertainment in the sense of “engagement” |
| 21:57 |
Helenah |
Ah |
| 21:58 |
erle |
mrcheese yeah, let's get back to debating if my new hardware counts actually as “old hardware” because it does not outperform someone's thinkpad! :P |
| 21:58 |
Helenah |
:D |
| 22:00 |
erle |
Helenah if you ever visit berlin, i can give you obscene (mostly queer) stickers though. i have printed a bunch. the most popular is “Atomkrieg? Ja, bitte!” |
| 22:01 |
erle |
styled after “Atomkraft? Nein, danke!”, if you know that |
| 22:05 |
Helenah |
erle: Ja, danke |
| 22:13 |
MTDiscord |
<nathan4220776> The Germans really want to go for round 3, huh? ( ͡° ͜ʖ ͡°) |
| 22:14 |
erle |
nathan yeah but this time on the side of the good guys, you know atonement and all that |
| 22:15 |
erle |
to be a bit on topic again: are there actually luanti games with interesting explosives? |
| 22:15 |
erle |
like, not just “delete/dig nodes” |
| 22:17 |
mrcheese |
idk |
| 22:20 |
cheapie |
<Helenah> How insensitive of me, you're transgender? ← This is IRC, are you really surprised? :P |
| 22:20 |
cheapie |
(I am too, for the record) |
| 22:24 |
Helenah |
cheapie: More like I went through a shock because of what I said... anyway... brb |
| 22:25 |
erle |
ｃａｔ ｅａｒｓ ｍｕｓｔ ｂｅ ｗｏｒｎ ｂｅｙｏｎｄ ｔｈｉｓ ｐｏｉｎｔ |
| 22:25 |
erle |
dear people, does anyone remember the bangers “luola” and “liero”? |
| 22:32 |
|
panwolfram joined #luanti |
| 22:39 |
luatic |
erle: a strictly CFG-based file format can essentially only support tree-like data structures well. |
| 22:40 |
luatic |
to be a good model file format, gltf effectively needs a DAG when it comes to various "object references". e.g. we want to be able to reference the same primitive multiple times. we want to be able to reference the same material multiple times. and so on |
| 22:40 |
rubenwardy |
I like Berlin |
| 22:40 |
rubenwardy |
all the pipes remind me of a windows 95 screensaver |
| 22:40 |
erle |
LMAO construction works indeed do have a lot of piping |
| 22:41 |
MinetestBot |
[git] Desour -> luanti-org/luanti: Make proper use of override in server.h 20f8eb9 https://github.com/luanti-org/luanti/commit/20f8eb9a6c7a518771ae677c42b535966cef6337 (2025-09-21T22:40:30Z) |
| 22:42 |
erle |
luatic you are thinking of an SVG-like <defs> data structure, right? |
| 22:42 |
rubenwardy |
I love Romeo und Romeo in Schonberg, wish we had somewhere like that in London. Soho is just too expensive and busy for it to exist |
| 22:43 |
erle |
Schöneberg? |
| 22:43 |
luatic |
erle: that is similar, yes |
| 22:43 |
erle |
there is a burger restaurant that is there and it is called “Schöne Burger” (i.e. “beautiful burgers”) |
| 22:44 |
rubenwardy |
Sorry yes. Schöneberg |
| 22:45 |
erle |
luatic “maybe json is the wrong format then” (which is arguably not entirely right, given that json schema itself has references) |
| 22:45 |
erle |
luatic but also, maybe, just maybe. the idea of gltf to use indices instead of object names is something that came out of “i want this struct serialized, DO IT” |
| 22:46 |
erle |
luatic at least to me gltf json serialization looks a lot more like “we want to expose the internal data structure mostly-verbatim” than “we want this easy to parse and validate”. what do you think? |
| 22:48 |
luatic |
indices are a bit user-unfriendly (vs names) but implementor-friendly 🤷 |
| 22:48 |
erle |
as i said |
| 22:49 |
erle |
i believe the issue is that implementors will always want a ransport format that could provide binary data for GPUs directly, without any “unnecessary” parsing |
| 22:50 |
erle |
transport |
| 22:52 |
luatic |
can't say i'm a big fan of how gltf did that bit (though you most probably want some sort of dense binary data format for vertex buffers) |
| 23:05 |
|
Eragon joined #luanti |
| 23:34 |
|
Thermoriax joined #luanti |
| 23:39 |
Helenah |
Does anyone know where I can find a guide that explains the different technic machine upgrades? (The ones you place into the slots of technic machines). |
| 23:41 |
Helenah |
Also, something seems to drain my battery boxes so they never get a charge. |
| 23:42 |
Helenah |
Ah, it fluctuates when the day is shining, I have solar power. |
| 23:44 |
erle |
Helenah it's not luanti, but have you ever checked out colobot? :3 |
| 23:46 |
Helenah |
No |
| 23:46 |
|
mrcheese joined #luanti |
| 23:46 |
user2_ |
yeah this isnt very related to luanti at all- |
| 23:46 |
mrcheese |
yeah XD |
| 23:47 |
mrcheese |
it almost never is lo |
| 23:47 |
mrcheese |
l |
| 23:48 |
Helenah |
Wait, what? |
| 23:48 |
Helenah |
Modding discussion is allowed here, user2_ |
| 23:48 |
mrcheese |
yes it is- |
| 23:49 |
user2_ |
yes, but most of this isn't modding afaik- |
| 23:49 |
mrcheese |
yea |
| 23:49 |
user2_ |
such as how it was fonts a few hours ago- |
| 23:50 |
Helenah |
Damnit! |
| 23:51 |
Helenah |
The battery box's charge level fluctuates so as long as it's day time, this is annoying, I tried breaking and replacing the switch stations and everything. |
| 23:56 |
Helenah |
OH, some energy is used by the box itself! Right, I get it now. :D |
| 23:59 |
Helenah |
That feature seems a bit weird. :D |
