Minetest logo

IRC log for #minetest-dev, 2022-05-04

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

All times shown according to UTC.

Time Nick Message
00:37 MTDiscord <MisterE> sfan5, I would like to ask for a way to remove dynamically sent media from the client. There are a bunch of cool applications for dynamic media possible if only they would not fill up the client's media cache.
00:53 basicer joined #minetest-dev
00:58 Taoki joined #minetest-dev
01:03 basicer I saw in the IRC logs there was some breif interest in using Luau in Minetest.  Did anything ever come of that?
01:05 MTDiscord <Jonathon> someone got it semi working, it was more for fun, rather than anything serious
01:07 basicer I made some minor minetest contributions back in the day, and I currently work on Luau, so it was cool to see the crossover.
01:08 v-rob joined #minetest-dev
01:19 erle MisterE what kind of ephemeral media would you explicitly want to remove? isn't it good enough to have it until end of session? are you trying to do really bad video streaming?
01:21 basicer Did server sent mods ever get decided on or implemented?
01:22 MTDiscord <Jonathon> no
01:32 v-rob joined #minetest-dev
01:50 MTDiscord <exe_virus> Though, I do know that c55 is interested/on board for the physics client side Lua stuff, so it's in the cards :)
02:04 erle exe_virus everyone is interested, but the only things anyone knows to implement securely and without performance issues are a) separately downloaded or integrated CSMs (dragonfire/waspsaliva/cheatdb) b) declarative solutions – and i would be extremely surprised if physics was not a declarative thing.
02:05 erle basicer have fun https://git.minetest.land/luk3yx/sscsm
02:06 MTDiscord <Jonathon> thats one of two options, but the only one thats being used in a engine
02:12 MTDiscord <Jonathon> https://github.com/GreenXenith/sscsm_ssm and https://github.com/GreenXenith/sscsm_csm
02:14 MTDiscord <LandarVargan> Note that those aren't necessarily implemented securely either
03:04 basicer Luckily I dont have a horse in the SSCSM race anymore.
03:13 MTDiscord <Jonathon> checking back in for something specific, or just in general?
03:18 basicer Just curious.  I was mostly interested in what came of the interest in Luau.
03:26 MTDiscord <FatalError> SSCSMs basically will never become a thing in this current climate tbh
03:41 basicer Im just happy to see the project still going 4 years later.
04:00 MTDiscord joined #minetest-dev
04:25 erle basicer what did you want to do using SSCSMs?
04:25 MTDiscord <MisterE> Erle basically, yes
04:25 MTDiscord <MisterE> There are many applications for really bad streaming
04:26 MTDiscord <MisterE> Its called moving pictures
04:27 v-rob joined #minetest-dev
04:28 MTDiscord <MisterE> Also, other things like billboards that can cycle thru an unlimited number of different ads, bgm that can be dj'd
04:29 MTDiscord <MisterE> Having a cctv mod would be possible
04:30 MTDiscord <MisterE> Having a tv channel mod ingame would be possible
04:31 MTDiscord <MisterE> Basically, anything that you could send/generate an unlimited number of small textures, which would be used once and then should be thrown away
04:37 erle MisterE[m] are texture modifiers are garbage collected? if so, i'd suggest to invent one that is less of a fractal of bad design than the “[PNG” one and utilize that. like plain PBM or so.
04:37 erle i bet they aren't though
04:38 erle MisterE[m] do you have any proof-of-concepts that run out of memory rn?
04:45 basicer erle: I dont want to do anything anymore.  At the time I was working for a company that made games for kids on a Raspberry Pi, and we wanted to have an educational version of minecraft for them.  If the server could send mods, it made our update story much easier.
04:54 erle basicer what update story?
04:55 basicer We had CSMs that were pretty complicated, so it would have been nice if the server could have sent those down instead of us needing to have scripts to update them from our srevers.
06:16 calcul0n joined #minetest-dev
06:21 v-rob joined #minetest-dev
06:26 talenone joined #minetest-dev
06:29 talenone is this channel active ?
06:48 MTDiscord <ROllerozxa> it's active when there's something to talk about
06:51 talenone joined #minetest-dev
07:18 tuxedoedpenguin[ joined #minetest-dev
07:24 Baytuch joined #minetest-dev
07:28 Zughy[m] joined #minetest-dev
07:28 Zughy[m] <MTDiscord> "<FatalError> SSCSMs basically..." <- considering how PRs are going down (subtract possible close + one approval PRs and you'll reach something like 80 PRs), maybe SSCSM discussion is not that far away. Just, one step at a time
07:29 programmerjake joined #minetest-dev
07:29 tochigi joined #minetest-dev
07:29 freshreplicant[m joined #minetest-dev
07:29 MisterE[m] joined #minetest-dev
07:29 _Zaizen_[m] joined #minetest-dev
07:29 MaxStirner[m] joined #minetest-dev
07:35 sfan5 @MisterE this requires garbage collection of textures at which point you no longer need an explicit way to remove sent textures
07:45 sfan5 paradust: does your fix for #11852 perhaps also have something to do with #11410?
07:45 ShadowBot https://github.com/minetest/minetest/issues/11852 -- Parts of slime entity model are not rendered anymore when shaders are turned on (regression from 5.4.1)
07:45 ShadowBot https://github.com/minetest/minetest/issues/11410 -- Flower pots sometimes get invisible in Hades Revisited
08:11 proller joined #minetest-dev
09:12 erle paradust thank you for figuring #11852 out. the answer for “why is this model like that” is most likely “you can make it look cooler using different textures”
09:12 ShadowBot https://github.com/minetest/minetest/issues/11852 -- Parts of slime entity model are not rendered anymore when shaders are turned on (regression from 5.4.1)
09:14 erle paradust like, i am pretty sure that there is one texture set in which the slime body is a transparent cube in a smaller cube and the eyes and mouth are just black cubes.
09:14 erle i just don't know its name
09:15 erle paradust with the other texture set it looks somewhat similar to that i think? https://static.wikia.nocookie.net/minecraft_gamepedia/images/d/dd/Slime_JE3_BE2.png/revision/latest?cb=20191230025505&amp;sa=X&amp;ved=2ahUKEwiWqNP_wMX3AhUSSfEDHXI-DbAQ_B16BAgEEAI
09:49 Fixer joined #minetest-dev
10:07 Taoki joined #minetest-dev
10:15 Zughy[m] Can I please have some core dev take on https://github.com/minetest/minetest/issues/10294#issuecomment-1117144499 ?
10:16 HuguesRoss joined #minetest-dev
10:31 erle Zughy[m] what exactly would be the bug in your minigame mod?
10:31 erle Zughy[m] also i bet rubenwardy can grep for how common it is to use this deliberately
10:41 MTDiscord <MisterE> Erle, I'm not sure. My dynamic_bgm mod, perhaps. Not sure, it would only affect clients with low memory such as mobile
10:41 erle why would anyone use noclip that way
11:02 MTDiscord <luatic> erle: we'd need your semantic grep here ;)
11:02 Zughy[m] erle: there is no bug, but there will be if collisionbox = {0} starts returning a one point collision again
11:03 Zughy[m] and I'd need physical = false to also be implemented
11:17 erle luatic you can write the semgrep rule as good as i can
11:27 sfan5 Zughy[m]: not changing that behaviour again sounds good to me, but physical = false should be implemented too
11:28 HuguesRoss Merging #11920 in 15
11:28 ShadowBot https://github.com/minetest/minetest/issues/11920 -- Add vector.combine by appgurueu
11:30 TheCoffeMaker joined #minetest-dev
12:40 proller joined #minetest-dev
13:55 paradust sfan5: Seems plausible that it does
13:56 paradust erle: Even with the simple fix, there's no way for the shader to render transparency inside transparency like that
13:57 erle paradust then ig i misremember how it looked.
13:57 paradust I believe this requires rendering the submeshes further from the camera first
13:58 erle have you tried it?
13:58 erle at least about the eyes thing i am pretty sure, that they are black cubes with another texture set.
13:58 paradust nope
13:58 paradust the eyes would probably be fine, since they are darker and not transaprent. but I suspect the inner cube would not show up.
13:59 paradust actually, it might be possible to fix this without doing anything fancy. maybe just by getting rid of the depth test
13:59 erle i mean, you can see how the textures look
14:00 erle there is enough space to just paint something in it
14:02 Zughy[m] #12269 has a bad title and it doesn't belong in MT repo, but I can't fix neither (if we exclude the "just close it" option)
14:02 ShadowBot https://github.com/minetest/minetest/issues/12269 -- minecart
14:16 MTDiscord <savilli> you can put Invalid tag
14:21 Zughy[m] oh, nice idea, thanks savilli
14:34 MTDiscord <ROllerozxa> minecart
14:38 Toothless joined #minetest-dev
15:28 v-rob joined #minetest-dev
15:50 appguru joined #minetest-dev
15:51 appguru Zughy: Could you please label #12270  as bugfix and #11904 as bug?
15:51 ShadowBot https://github.com/minetest/minetest/issues/12270 -- HUD: Update selection mesh every frame by appgurueu
15:51 ShadowBot https://github.com/minetest/minetest/issues/11904 -- Minetest doesn't update entity selectionbox while selected
15:53 appguru Zughy[m]
15:53 Zughy[m] done everything
15:53 appguru thanks :)
15:54 Zughy[m] I get notified if anyone writes just my name, it's a Matrix feature. No need to tag me :D yw
15:54 appguru BTW: Do you think a HUD tag might be useful?
15:54 appguru Considering there's a formspec tag as well
15:54 Zughy[m] there is UI/UX
15:55 sfan5 what makes UI/UX different that it doesn't fit into any of the @ categories (like "@ Mainmenu")?
15:55 sfan5 not that I disagree I'd just like to know whether you have given it thought
15:56 rubenwardy agreed
15:56 fluxionary joined #minetest-dev
15:56 rubenwardy it feels redundant with formspec
15:56 rubenwardy I'd maybe merge formspec into UI/UX
15:57 rubenwardy UX does include other things, like controls and general ease-of-use
15:57 Zughy[m] it's mostly UX, so an issue that complains about user experience (accessibility, touch experience etc)
15:58 Zughy[m] sfan5: #3022 #3255 and #11058 now belong to MT fork of irrlicht. They should be either moved or closed
15:58 ShadowBot https://github.com/minetest/minetest/issues/3022 -- Input Method Editor (IME) Support
15:58 ShadowBot https://github.com/minetest/minetest/issues/3255 -- Some Irrlicht messages bypass the Minetest log system
15:58 ShadowBot https://github.com/minetest/minetest/issues/11058 -- "M" key not working on MacOS X french keyboard
15:58 rubenwardy Zughy[m]: how much of this is done now #2729 ?
15:58 ShadowBot https://github.com/minetest/minetest/issues/2729 -- Sky additions/changes
15:59 sfan5 Zughy[m]: I think they should be reopened on the irrlicht repo if still applicable instead of moving
16:00 Zughy[m] rubenwardy: just the layering, which is better explained in #11366 anyway. I'll close the one from Kenney
16:00 ShadowBot https://github.com/minetest/minetest/issues/11366 -- Sky layer API
16:01 rubenwardy cool! Make sure to add a comment
16:03 sfan5 github says someone mentioned me in #12261 4 minutes ago but I don't see anything
16:03 sfan5 weird
16:03 ShadowBot https://github.com/minetest/minetest/issues/12261 -- [no squash] Optimize JSON string (de)serialization routines by sfan5
16:03 Zughy[m] sfan5: I'll put a comment telling people to reopen there then, closing
16:05 rubenwardy Do we provide official MacOS builds?
16:05 rubenwardy looks like no
16:05 rubenwardy context: #3596
16:05 ShadowBot https://github.com/minetest/minetest/issues/3596 -- Making Minetest easy to install on all operating systems
16:09 sfan5 travis artifacts are the closest we have
16:09 sfan5 but software won't run unless you somehow let apple sign it first
16:09 sfan5 so those don't work(?)
16:10 sfan5 anyway we'd need a person with a macbook to sit down for a week to actually make macos stuff work properly
16:11 Zughy[m] I wonder how many people with a macbook there are in the MT community
16:14 erle jordach?
16:15 erle surely the apple users can just appeal to the maintainers of their distribution
16:15 MTDiscord <Warr1024> I have an intel macbook, though it's not my primary system for MTing
16:17 MTDiscord <Warr1024> "Connect to a server with any of a variety of devices and get a reasonably consistent experience" is one of the things I really like about MT.
16:17 MTDiscord <Warr1024> I'd play MC if I wanted "we have interoperability at home"
16:18 erle sick burn hehe
16:18 erle i sometimes read mojangs issue tracker. apparently they have – contrary to minetest – not figured out how to not ruin user worlds when migrating them to newer versions. hilarious.
16:19 erle sfan5 how does homebrew solve the apple landlord problem?
16:20 erle context: i remember using a mac (for work, i complained until i got a thinkpad) and homebrew was a package manager that installed stuff from the internet. and i doubt all of that had been properly signed.
16:22 MTDiscord <Warr1024> I use homebrew, and I think it has its own signature system, which of course doesn't interop with the Apple signing stuff so you still get copious warnings when you run stuff you installed via brew the first time.
16:22 Alias joined #minetest-dev
16:22 MTDiscord <Warr1024> I actually like using a mac for work because it's easier not to mix business and pleasure when they're on different platforms.
16:23 MTDiscord <Warr1024> IIRC there are some builds for mac on the forums, but I think I use the homebrew version and it's adequate, though there are a number of random annoyances to it
16:23 MTDiscord <Warr1024> like I sound like a dunce when I type in chat on the macbook because for some reason the apostrophe key doesn't work at all so I type "youre" a lot there.
16:29 Evergreen joined #minetest-dev
16:30 sfan5 anyway, reminder that #11545 needs another review
16:30 ShadowBot https://github.com/minetest/minetest/issues/11545 -- Animated particlespawners by velartrill
16:32 erle uh, does the “take a step towards parity with the particle systems of other game engines” include getting rid of a bazillion drawcalls?
16:32 erle doesn't seem so, well
16:34 rubenwardy that's my view, feels like building on unstable foundations
16:35 MTDiscord <luatic> MT has always been an unstable foundation.
16:35 sfan5 neither of you seem to have read https://github.com/minetest/minetest/issues/11505#issuecomment-890941573
16:35 erle also these features are nice enough that people will immediately depend on that
16:37 MTDiscord <luatic> I agree, this is to be viewed independent of perf issues
16:37 rubenwardy it's not a separate concern when the code is intertwinned. That's a rather idealistic comment
16:38 MTDiscord <luatic> Pretty sure velartrill hardly touched the rendering code (only in relation to the blending changes, which hecks' comment predates)
16:40 sfan5 it's not idealistic, the reasoning is explained
16:40 erle i only know that the irrlicht particle system already has affectors. and i still think the easy way out would be to use that (once someone figures out how to render rain with it, or have a separate path for that maybe). i doubt it is that easy after more features are bolted on.
16:41 erle i guess particles are, after all, a bunch of scene nodes with billboards
16:42 erle is velartrill on irc?
16:42 erle (i can't see discord users)
16:44 erle “Performance wise, I don't think you could make particles any worse even if you tried. =]” haha i remember the issue now
16:46 erle does devtest have a particle stress test? i mean particle lag happens in real games all the time, but is there anything? (i haven't seen anything)
16:48 MTDiscord <luatic> velartrill is not on Discord
17:46 sfan5 not to mention that after the contributor had presented a proposal, discussed it with devs, received a go ahead, did all the coding work and received one approval it's not the right point in time to say "uh no actually we didn't like this from the start"
17:53 MTDiscord <luatic> velartrill did a good job
17:59 rubenwardy huh wtf
17:59 rubenwardy trying to find the issue about disabling game mods
17:59 rubenwardy another closed issue mentions it, was closed as duplicate, and links to https://github.com/minetest/minetest/issues/8622/
17:59 rubenwardy but that doesn't exist
18:00 rubenwardy was it GDPR'd
18:06 wsor stuck between two kidscode prs of all things
18:07 rubenwardy issues should be owned by the project, not the user. They shouldn't be able to just delete issues
18:07 rubenwardy although I guess maybe it was closed
18:14 sfan5 there's no direct way I'm aware of by which users can delete issues
18:14 sfan5 deleting their account will reown issues to 'ghost'
18:20 Zughy[m] I know some changes have been made to the CI recently, so is this thing still relevant and/or wanted? #11667
18:20 ShadowBot https://github.com/minetest/minetest/issues/11667 -- Github CI: Reduce compilation times with ccache
18:30 MTDiscord <luatic> Zughy: yes, it is still relevant; sfan5's fixes are just fixes (and upgrades) AFAIK and not performance improvements
18:30 MTDiscord <luatic> Wasn't there another PR that suggested using multiple cores?
18:34 sfan5 the github runners have only two cores and our scripts already auto-detect the number of cores
18:36 paradust are we building -O0 and without debug info?
18:37 sfan5 -O0 and with debug info, mostly
18:40 Krock will merge #12247 #12256 and #12215 in 15 minutes
18:40 ShadowBot https://github.com/minetest/minetest/issues/12247 -- Make logging cost free when there is no output target by paradust7
18:40 ShadowBot https://github.com/minetest/minetest/issues/12256 -- guiScalingFilter: Fix most memory leaks by SmallJoker
18:40 ShadowBot https://github.com/minetest/minetest/issues/12215 -- hud_get: Return precision field for waypoints by appgurueu
18:54 Krock merging
18:55 Krock done
18:57 MTDiscord <luatic> Anyone willing to take a look at #12270? It fixes a pretty ancient issue.
18:57 ShadowBot https://github.com/minetest/minetest/issues/12270 -- HUD: Update selection mesh every frame by appgurueu
19:22 sfan5 how expensive is updateSelectionMesh?
19:24 MTDiscord <luatic> would you like me to profile it?
19:28 MTDiscord <luatic> code-wise all it's doing is creating a halo mesh - literally a single cube
19:28 MTDiscord <luatic> should be very fast
19:29 sfan5 profile it just in case
19:30 MTDiscord <luatic> Can I just put a ScopeProfiler in the func scope and it will automagically work?
19:31 sfan5 yes
19:32 sfan5 it measures in milliseconds though which probably rounds to zero
19:33 sfan5 it drops and recreates the mesh though, this will be bad because it will require it to be reuploaded every frame
19:33 MTDiscord <luatic> https://cdn.discordapp.com/attachments/747163566800633906/971494878045438002/Bildschirmfoto_von_2022-05-04_21-33-30.png
19:34 sfan5 (and this might not show up in the profiler)
19:34 MTDiscord <luatic> true
19:34 MTDiscord <luatic> but this is already the current worst-case
19:34 MTDiscord <luatic> should I include a check whether the mesh has changed?
19:34 sfan5 is it?
19:34 MTDiscord <luatic> sfan5: if you fly around point at things (e.g. in a tunnel), your pointed thing is already constantly changing
19:35 sfan5 but not every frame
19:35 MTDiscord <luatic> if you're fast enough...
19:35 sfan5 well ok
19:35 sfan5 but with this you'd be making the average case the worst case
19:36 MTDiscord <luatic> indeed
19:37 sfan5 hm I can't think of a good way to detect updates
19:37 sfan5 so I guess having it not touch the mesh if it's equal would be a good way
19:41 Zughy[m] I proudly inform you that issues went down from 1104 to 1046 in about 24 hours thanks to Rubenwardy and me. Now the counter doesn't say 1.1k anymore, just 1k; I hope to bring them below 1000 eventually
19:42 MTDiscord <ROllerozxa> yay, I was just gonna mention how much the open issue amount has shrunk the last day or so but you beat me to it :P
19:46 Zughy[m] <3
19:47 Zughy[m] if anyone wants to test unconfirmed bugs and then just ping me when you can confirm one, please do and I'll change the labels
19:56 sfan5 appguru: actually you can probably locate the code that sets m_selection_boxes and have it set a changed flag
19:57 sfan5 https://github.com/minetest/minetest/blob/a8707158a5be8c604fdb028a9c5f68ce5804016e/src/client/hud.cpp#L857-L865 haha never mind this is terribly inefficient regardless of how often the mesh is changed
19:59 behalebabo joined #minetest-dev
20:06 MTDiscord <luatic> sfan5: I take it I shouldn't implement the checks in the end?
20:07 MTDiscord <luatic> (yes, this already generates a mesh & sends it to the GPU each frame anyways)
20:08 sfan5 you can skip that yes
20:31 MTDiscord <luatic> sfan5: will shutdowns wait for async threads to finish?
20:34 sfan5 yes
20:35 sfan5 it's kinda hard not to
20:37 v-rob joined #minetest-dev
20:38 appguru good good, that allows me to implement async file writing using them, right?
20:44 Guest4738 joined #minetest-dev
20:58 sfan5 yes
21:14 proller joined #minetest-dev
21:37 MTDiscord <x2048> Merging #12262 in 5m
21:37 ShadowBot https://github.com/minetest/minetest/issues/12262 -- Avoid rendering invisible faces of simple nodeboxes by x2048
21:38 MTDiscord <Warr1024> NICE
21:38 MTDiscord <Warr1024> and also SHIT
21:39 MTDiscord <Warr1024> now this means I have to undo some of my nodebox->mesh optimizations when they become pessimizations.
21:46 MTDiscord <x2048> Merged now
21:47 rubenwardy merging #11176 in 5
21:47 ShadowBot https://github.com/minetest/minetest/issues/11176 -- Fix Windows Visual Studio actions by LoneWolfHT
21:47 MTDiscord <x2048> @Warr1024 well, NodeCore will only get better 🙂
21:47 rubenwardy also #12115 is pretty trivial
21:48 ShadowBot https://github.com/minetest/minetest/issues/12115 -- Default chat clickable weblinks by Froggo8311
21:49 MTDiscord <Warr1024> x2048, I'd love to see more geometry optimizations like that.  If you're interested in something along the same theme, it'd be nice to generate separate more-optimized wieldmeshes for different images based on alpha-clip-mask as well as resolution, instead of the generic one with tons of hidden faces.  Dunno the specific # of that one though...
21:50 Toothless joined #minetest-dev
21:51 MTDiscord <Warr1024> #3069 omg wow, this one is ancient...
21:51 ShadowBot https://github.com/minetest/minetest/issues/3069 -- Many wieldmesh polygons
21:51 MTDiscord <Warr1024> Speaking of NodeCore, this one is possibly a BIG deal for NodeCore.
21:51 Toothless Is there a way to force remove entities, like removing an train from advtrains.
21:55 MTDiscord <x2048> Toothless: it might be best to search/ask in the forum, like, here : https://forum.minetest.net/viewtopic.php?f=11&amp;t=14726
22:34 panwolfram joined #minetest-dev
23:08 Zughy[m] Pretty trivial but "Maintenance" label should be "@ Maintenance" to be coherent with the rest of the macrocategories (at least, the colour is the same)
23:09 rubenwardy maintenance isn't an area though, it's a thing you do with any area
23:10 rubenwardy @ is supposed to be a location in the code, a subsystem
23:13 AliasAlreadyTake joined #minetest-dev
23:33 Guest8471 joined #minetest-dev
23:35 Toothless joined #minetest-dev
23:58 Guest2893 joined #minetest-dev

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