Minetest logo

IRC log for #minetest-dev, 2022-08-02

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:02 Baytuch joined #minetest-dev
00:10 Zughy[m] also, anyone with Windows 11, if you can test this (it takes 20 seconds) #12627
00:10 ShadowBot https://github.com/minetest/minetest/issues/12627 -- Crash when trying to change font while clicking the browse button (VIDEO)
00:14 MTDiscord <Jordach> it's installed to a read only directory
00:14 MTDiscord <Jordach> Program Files and it's x86 cousin are read only to users
00:15 MTDiscord <Jordach> Minetest is expecting to write there
00:15 MTDiscord <Jordach> this will also go for the conf afaik
00:28 MTDiscord <Jordach> implemented minetest's scale, offset and spread
00:33 Warr1024 joined #minetest-dev
00:52 Warr1024 joined #minetest-dev
01:07 Warr1024 joined #minetest-dev
01:34 Warr1024 joined #minetest-dev
01:39 Taoki joined #minetest-dev
02:17 Warr1024 joined #minetest-dev
02:28 YuGiOhJCJ joined #minetest-dev
02:40 Baytuch joined #minetest-dev
03:14 Baytuch_2 joined #minetest-dev
03:30 YuGiOhJCJ joined #minetest-dev
04:00 MTDiscord joined #minetest-dev
04:07 MTDiscord <Jordach> ah right perlin doesn't like integers
04:15 MTDiscord <Jordach> more research required
05:37 calcul0n joined #minetest-dev
06:38 Baytuch joined #minetest-dev
06:55 Baytuch joined #minetest-dev
07:14 MTDiscord <ROllerozxa> there should be some kind of check in minetest.register_node if the tiles key is a string instead of a table, since some new modders think that will work for a single texture node and wonder why it's not showing up
07:35 Baytuch_2 joined #minetest-dev
09:05 appguru joined #minetest-dev
09:14 MTDiscord <luatic> rubenwardy: Current pattern for readable name is "%(([^%)]*)%)"  -- readable name, so the answer is that they may have open, but not closing brackets :P; to allow for matching brackets, you'd have to simply replace this mess with (%b())
09:15 rubenwardy Context?
09:15 rubenwardy Oh setting types
09:44 olliy joined #minetest-dev
09:50 sfan5 merging #12622, #12131 in 5m
09:50 ShadowBot https://github.com/minetest/minetest/issues/12622 -- MeshUpdateQueue fixes by sfan5
09:50 ShadowBot https://github.com/minetest/minetest/issues/12131 -- Add `minetest.settings` to CSM API and allow CSMs to provide `settingtypes.txt` by AFCMS
10:00 MTDiscord <x2048> I've narrowed down #12621 to particles - https://user-images.githubusercontent.com/4933697/182347310-4eb74743-d958-4699-b5b5-43e71b4a86a5.png
10:00 ShadowBot https://github.com/minetest/minetest/issues/12621 -- Random freezes in Mineclone 2 0.77
10:02 kilbith joined #minetest-dev
10:02 sfan5 interesting
10:12 * rubenwardy hopes this is the spark that gets x2048 to implement particle batching
10:21 MTDiscord <x2048> From what I see the problem is collision detection
11:26 Zughy[m] Again, I need more powers on GH to properly manage all the issues and PRs. Practical example: renaming #12633
11:26 Zughy[m] I'm kind of tired of having to ask every time to some core dev to do this and that
11:26 ShadowBot https://github.com/minetest/minetest/issues/12633 -- FPS on Android works great, but linux, not so great
11:31 MTDiscord <ROllerozxa> that issue could just be closed altogether. first part mostly likely would be a duplicate of some other issue talking about mouse support on Android, second part is an upstream issue about how chrome os's Linux container has subpar graphics rendering
11:44 ROllerozxa joined #minetest-dev
11:44 HuguesRoss1 joined #minetest-dev
11:54 cranezhou joined #minetest-dev
12:13 HuguesRoss6 joined #minetest-dev
12:51 Warr1024 joined #minetest-dev
13:19 Warr1024 joined #minetest-dev
13:42 Fixer joined #minetest-dev
13:53 Warr1024 joined #minetest-dev
14:07 Baytuch joined #minetest-dev
14:28 Wuzzy joined #minetest-dev
14:38 Yad joined #minetest-dev
14:48 appguru joined #minetest-dev
15:12 Baytuch joined #minetest-dev
15:18 Baytuch joined #minetest-dev
15:38 Baytuch joined #minetest-dev
15:39 Warr1024 joined #minetest-dev
15:49 MTDiscord <Ronoaldo> Is there any hints on how to run a development debug session on VS Code?
15:49 Alias joined #minetest-dev
15:52 MTDiscord <Jordach> gdb and valgrind are that way
15:52 MTDiscord <Jordach> If you’re on windows then use MSVC itself as a debugger
15:54 MTDiscord <Ronoaldo> got it - I'll try to run it from Linux with gdb
15:55 Baytuch joined #minetest-dev
16:06 Zughy[m] nice, looks like we're basically ready for 5.6
16:27 Toadally joined #minetest-dev
16:27 Toadally left #minetest-dev
16:29 rubenwardy !title https://twitter.com/rubenwardy/status/1554502669955436544
16:29 ShadowBot rubenwardy: Error: That URL appears to have no HTML title.
16:29 rubenwardy Minetest on the steam deck
16:29 rubenwardy Once joysticks are enabled, moving, rotating, digging, and placing all work. It's not possible to open the pause menu, inventory, or close from the main menu when in fullscreen
16:30 Toadally joined #minetest-dev
16:31 Toadally @kilbith hey, I heard you're working on a Camera API. There's some functionality I wanted to add to the camera/other objects displaying on the screen, what's the status on that for you, and will I be able to fork it to add on?
16:33 kilbith it has been delayed by my hierarchy due to more urgent stuff, but the development will resume mid or end of August.
16:34 kilbith I'd say I'm midway with this API
16:34 kilbith diff is circa +500-3
16:38 olliy joined #minetest-dev
16:45 Toadally <kilbith> where can I fork it? assuming that's probably the best way to do it right now
16:47 kilbith ?
16:47 kilbith why do you want to fork it?
16:48 kilbith the code is not even submitted to MT
16:48 Toadally I was told to refer to you since i'm looking into making some camera changes
17:00 MTDiscord <MisterE> since you are both doing work on the camera, maybe you two should collaborate...
17:34 Alias joined #minetest-dev
17:37 MTDiscord <Ronoaldo> @rubenwardy that is awesome 🙂 glad to hear it is almost working on steam deck
18:10 Fixer_ joined #minetest-dev
18:39 MTDiscord <x2048> @Ronoaldo I use cpptools extension and almost default launch.json https://gist.github.com/x2048/e6fdc526126a997c8b9d708f54375eee
18:42 MTDiscord <Ronoaldo> @x2048 thanks for sharing!!! I'll test it out
18:43 MTDiscord <x2048> One caveat is that it does not build on run, I build from console, then F5 to debug.
18:45 MTDiscord <Ronoaldo> oh - that is where I made a mistake then - I just hit Debug assuming it would magically build lol
18:45 MTDiscord <Ronoaldo> what flags you build with?
18:45 mrkubax10 joined #minetest-dev
18:46 mrkubax10 x2048: you can create build task to compile and then specify "preLaunchTask" in launch.json
18:49 MTDiscord <x2048> I know I can configure magic in VSCode, I just don't care. ^R make in bash does its job.
18:52 MTDiscord <x2048> @Ronoaldo I configured with cmake as described in README.md, and then make -j$(nproc)
18:53 MTDiscord <x2048> To enable debugging you need to configure with cmake -DCMAKE_BUILD_TYPE=Debug . It does the rest.
18:59 Pexin joined #minetest-dev
18:59 MTDiscord <Ronoaldo> tks!
19:00 Yad joined #minetest-dev
19:00 MTDiscord <Ronoaldo> @mrkubax10 that's what I was looking for - so I just need to cmake && make on preLaunchTask?
19:01 mrkubax10 wait I will send example tasks.json and launch.json that I use
19:01 MTDiscord <x2048> Just make, no need to reconfigure all the time
19:01 MTDiscord <Ronoaldo> thank you
19:03 MTDiscord <Ronoaldo> Just to make sure I got the correct plugin, this is the one, right? https://code.visualstudio.com/docs/languages/cpp
19:05 MTDiscord <x2048> Looks correct
19:06 mrkubax10 yes
19:07 mrkubax10 launch.json: https://termbin.com/63ly  tasks.json: https://termbin.com/4ql9
19:07 mrkubax10 this will require some configuration anyway
19:07 MTDiscord <Ronoaldo> sup! thank you again
19:18 Krock will merge #7629 in 15 minutes
19:18 ShadowBot https://github.com/minetest/minetest/issues/7629 -- Run Minetest update checker on startup by SmallJoker
19:32 Krock merging
19:34 Krock done
19:38 ROllerozxa #12637 should probably be done for 5.6.0
19:38 ShadowBot https://github.com/minetest/minetest/issues/12637 -- Disable dynamic shadow dropdown on OGLES2 by rollerozxa
19:38 ROllerozxa since dynamic shadows won't be functional on OGLES2
19:42 MTDiscord <Ronoaldo> thanks @x2048 and @mrkubax10 - it is working now! I also added the CMake extension which seems to help intelisense to figure out some stuff
19:42 mrkubax10 nice
19:44 MTDiscord <Ronoaldo> ok, now that I have the stuff working - do we have a "beginner friendly" list of issues somewhere?
19:45 rubenwardy there's a beginner friendly tah
19:45 rubenwardy label
19:47 Krock ROllerozxa: what about OGLES1?
19:48 rubenwardy that doesn't have shaders
19:48 ROllerozxa doesn't support shaders to begin with
19:49 Krock ah I see. the check for that is just a few lines higher up
19:49 Krock for safety reasons, I'd also disable the dynamic shadows setting in the else-case
19:50 Krock so if they were enabled by accident, be sure to disable them.
19:51 sfan5 release tomorrow then?
19:52 sfan5 assuming we can agree how to credit people by then
19:52 MTDiscord <Ronoaldo> yay! 🙂
20:03 MTDiscord <Ronoaldo> unit tests are executed as a make target, or are compiled into the program itself?
20:04 MTDiscord <Ronoaldo> nevermind, just found at the main.cpp file!
20:11 mrkubax10 left #minetest-dev
20:11 Baytuch joined #minetest-dev
20:17 Baytuch joined #minetest-dev
20:21 MTDiscord <x2048> sfan5: I think #12621 should be fixed before the release. MCL2 is unplayable.
20:21 ShadowBot https://github.com/minetest/minetest/issues/12621 -- Random freezes in Mineclone 2 0.77
20:25 MTDiscord <Ronoaldo> I got this right after a recent merge
20:25 MTDiscord <Ronoaldo> https://cdn.discordapp.com/attachments/747163566800633906/1004122730691559546/Screenshot_20220802_172450.png
20:27 MTDiscord <Ronoaldo> https://paste.debian.net/1249135/
20:27 sfan5 @x2048 if this is a case of "mcl2 spawns way too many particles because bad code" how are we supposed to fix it?
20:28 Baytuch joined #minetest-dev
20:29 ROllerozxa Krock: Done. Although keep in mind you can still weasel your way through by setting the video driver to OpenGL and being able to set it while still being under OGLES2, this even works for shaders itself on OGLES1 and is the cause of #12613
20:29 ShadowBot https://github.com/minetest/minetest/issues/12613 -- Shader failed to generate
20:31 MTDiscord <x2048> sfan5: mcl2 works without freezes on 5.5.1
20:34 sfan5 I see
20:34 sfan5 was the flamegraph not indicative of what's wrong?
20:37 MTDiscord <x2048> there were gaps in the main thread during the freezes. I have narrowed down the issue to particle collision detection so far.
20:42 MTDiscord <x2048> Double-checking the flamegraph....
20:42 MTDiscord <Warr1024> IIRC MCL2's particle abuse is a known issue, but even for games that use a pretty reasonable number of particles, there can be noticeable freezes because of the way MT handles particles.  Improving particle efficiency in the engine would be at least a nice way to meet in the middle.
20:43 MTDiscord <Warr1024> In NodeCore I've seen at least one case where I had a short (~100ms) animation, whose start coincided with a small burst of particles (like 20), but the lagspike caused by the particles was enough to mostly hide the animation.
20:45 MTDiscord <x2048> Here's a flamegraph of the main thread during a freeze: https://user-images.githubusercontent.com/4933697/182469866-64aec3b3-f07b-4c82-b1c4-e1f861274b1c.png
20:46 MTDiscord <Warr1024> Oooh, when you have particle collisions on, that forces the engine to constantly check whether there's a node there ... does that mean that by disabling particle collision checks we could get a big performance boost for stuff involving particles...?
20:49 MTDiscord <x2048> It collects nodeboxes in a cube [prev_pos, new_pos] and checks against all of them. There's place for improvement here.
20:51 sfan5 I'm sure there is but we don't want to rush in a collision detection refactor before release
20:51 sfan5 what I think makes the difference is that the old particle code for some reason performed *less* collisionMoveSimple() calls
20:51 sfan5 so that should be restored
20:55 MTDiscord <Warr1024> It does feel like gamedevs could be advised that colliding particles are significantly more expensive than non-colliding ones, and they should think very carefully about whether they REALLY need this flag.  I had been under the impression that particles were themselves fundamentally "slow" and hadn't realized that so much of it could be controlled by one flag.
20:56 MTDiscord <Warr1024> At least in my use-case, a lot of my particles are short-lived bursts that never move that far from their starting positions, so just turning off collisions would probably have a pretty minimal impact on the appearance, but being able to improve performance so cheaply would be nice.
20:57 MTDiscord <x2048> sfan5: Just compared 5.5.1 to 5.6.0, it's not the number of collisionMoveSimple().
20:57 MTDiscord <x2048> I suspect, it's the particle velocity/acceleration passed to collisionMoveSimple()
21:01 sfan5 if there's no movement this check might do it https://github.com/minetest/minetest/blob/2d8eac4e0a609acf7a26e59141e6c684fdb546d0/src/collision.cpp#L254-L256
21:03 sfan5 try replacing it with core::equals(speed_f->getLength(), 0.0f)
21:12 MTDiscord <x2048> no luck
21:13 sfan5 then I'd dump the paramters and compare between 5.5/5.6
21:13 MTDiscord <x2048> https://github.com/minetest/minetest/issues/12621#issuecomment-1203219375
21:13 MTDiscord <x2048> Logs of velocity and acceleration
21:14 sfan5 10**14 and infinity?
21:19 MTDiscord <x2048> That's what is sees
21:19 MTDiscord <x2048> *it sees
21:20 MTDiscord <x2048> I'm exploring now whether drag and jitter may come too high
21:20 MTDiscord <x2048> These are the new factors to velocity
21:30 Baytuch joined #minetest-dev
21:33 MTDiscord <x2048> Bounce
21:33 MTDiscord <x2048> 2022-08-02 23:32:02: INFO[Main]:   A_PARTICLES::max bounce  . . . . . . . . . .   46x  2.62335e+28
21:33 MTDiscord <x2048> 2022-08-02 23:32:02: INFO[Main]:   A_PARTICLES::min bounce  . . . . . . . . . .   46x  2.50153e+29
21:39 MTDiscord <x2048> sfan5: https://github.com/minetest/minetest/issues/12621#issuecomment-1203239464
21:40 rubenwardy static linting should be used to detect uninitialised values
21:40 rubenwardy nice spot though
21:41 natewrench joined #minetest-dev
21:45 MTDiscord <x2048> This is a trivial fix, I'm pushing it directly to master in a few minutes
21:55 MTDiscord <x2048> done
22:07 Zughy[m] reminder that anyone wants to help us test some unconfirmed bugs, you can have fun here: https://github.com/minetest/minetest/issues?q=is%3Aissue+is%3Aopen+label%3A%22Unconfirmed+bug%22+-label%3A%22Action+%2F+change+needed%22+-label%3A%22Possible+close%22
22:08 Zughy[m] *that if anyone
22:34 panwolfram joined #minetest-dev
22:58 Baytuch joined #minetest-dev
23:18 AliasAlreadyTake joined #minetest-dev
23:31 Baytuch joined #minetest-dev
23:39 Baytuch joined #minetest-dev

| Channels | #minetest-dev index | Today | | Google Search | Plaintext