Time Nick Message 00:59 kilbith hey hmmmm 00:59 hmmmm hi 01:00 kilbith whatcha doing here 01:00 hmmmm just hanging out, i miss you guys 01:01 rubenwardy oh hi 01:01 hmmmm hey 01:01 kilbith you do not miss nrz :] 01:01 hmmmm haha 01:01 hmmmm he probably mellowed out since then 01:02 hmmmm wasn't happy with the way commits were being shoveled in back then but at least i didn't let any of those vulns through 01:02 kilbith we have got another core-dev that you would love 01:02 hmmmm i hope the review process got better 01:04 kilbith tbh it's much less oppressive than we used to have when you were there 01:04 kilbith but the other hand the QA has lowered 01:04 kilbith +on 01:05 kilbith more regressions after merges, but it's fixed quick 01:05 kilbith it's more flexible than before 01:06 hmmmm every time i took a break from minetest and came back 6 months later, update to latest in master, and then try it out, there'd be a new stutter or lag 01:06 hmmmm lol 01:06 hmmmm drives me insane 01:06 kilbith you might take a look to what I tinkered with: https://www.youtube.com/channel/UCL3-3UWAMEjalYeGYzZvjxg/videos 01:06 rubenwardy maybe you're just forgetting the stutter or lag 01:06 kilbith I implemented ffmpeg in MT 01:06 rubenwardy :D 01:07 hmmmm like ilu guys but you gotta get better at profiling and optimizing the changes you made instead of just constantly adding in new changes 01:07 kilbith and we have got shadowmapping 01:07 hmmmm woow nice 01:07 rubenwardy it's hard to avoid regressions in a program with less than 13% unit test coverage. There's Lua unit tests and a couple of integ tests, but no real progress on regression tests 01:07 hmmmm lol i was working on that right before i got busy 01:07 rubenwardy I mean - shadowmapping is probably a good example of "new changes rather than profiling and optimising" 01:07 hmmmm yeah sorry for abandoning you 01:07 hmmmm i switched jobs, moved, etc. life happened 01:08 hmmmm i'm glad people are working on the hard stuff like shadowmapping 01:08 rubenwardy we have forked Irrlicht and have deleted like 90% of the code 01:08 rubenwardy it has allowed us to work on fixing input issues that were unfixable 01:08 hmmmm sweet 01:09 hmmmm yeah irrlicht used to be a huge pain point 01:09 rubenwardy well, I say we but I mean s*an and h*cks 01:09 hmmmm at some point after MT i started working on my own game, it's like a real time strategy with procedurally generated maps 01:09 kilbith yeah we have got hecks - who is literally a beast on the rendering dev 01:09 hmmmm i settled on irrlicht and got tired of it fairly quick 01:10 hmmmm that whole thing went nowhere fast 01:10 hmmmm oh i wonder how build a world ended up 01:10 hmmmm remember them 01:10 kilbith hmmmm: https://www.youtube.com/watch?v=uz1douFQcok 01:10 rubenwardy they pivoted to education and sold their company I believe? 01:11 hmmmm nice 01:11 hmmmm i mean the reflections 01:11 kilbith https://github.com/minetest/minetest/pull/9430 01:11 hmmmm didn't think we were gonna have any more shader work after RBA left us 01:11 hmmmm RIP 01:12 kilbith we trashed some of the RBA shaders eventually 01:12 rubenwardy hecks has filled his place, only 7 years later... 01:12 rubenwardy eh 01:12 hmmmm maybe they weren't the best, but at least they existed 01:13 hmmmm lol @ the ffmpeg commit 01:13 hmmmm so you got TV blocks or something now 01:14 kilbith I closed the PR 01:14 hmmmm aw 01:14 kilbith they said it's too big of a dependancy 01:14 hmmmm yeah it is, but i wouldn't complain about it if it were optional... 01:14 rubenwardy that was one issue, the other was that our media transfer system wouldn't have worked 01:14 hmmmm thing is, in a collaborative game like MT, you know everybody's gonna want to use it in their servers, and make it de facto mandatory 01:14 rubenwardy although - with dynamic media, it's more viable 01:15 rubenwardy (dynamic media = servers can send media at runtime, since v5.3) 01:15 kilbith yeah we can update media dynamically now 01:15 hmmmm 5.3?! 01:15 rubenwardy yeah, we went from v0.4.17 to v5.0 01:15 hmmmm ohh 01:15 rubenwardy dropping the 0. 01:15 hmmmm should probably went to 1 imho ohwell 01:15 hmmmm you pulled a SunOS 01:16 hmmmm how do you do the render to texture in game without causing significant lag? 01:16 MTDiscord I'm looking for some easy coding related stuff to get into. I've merged a couple PRs to the build system, so I have the slightest bit of experience with the PR system and the project structure. I don't mind code quality stuff and little issues that are quick fixes, but there's no system I'm aware of for an amateur contributor to find issues that are appropriate for him to work on. 01:17 hmmmm to be fair that's the issue with many of the open source projects, it takes a ton of prior context and skill to be useful much of the time 01:17 hmmmm "anybody can just contribute if they want!" is such a bald faced lie lol 01:18 rubenwardy yeah 01:18 rubenwardy we need more highly skilled contributors with time who can become core developers 01:18 MTDiscord Well, to my discredit, I did introduce more problems in my first PRs for me to subsequently fix, creating a cycle of constant opportunities for myself. 01:18 hmmmm my advice is to not look for easy stuff, but focus on one part of MT that you really want to make better, study up all you can on it, become an expert on it 01:19 MTDiscord Well, that's the build system and the code style, but neither align with the core devs goals afaik. XD 01:20 hmmmm most of the time the problems aren't difficult, per se, but they need somebody to understand how what currently exists works, why it's like that, how it interacts with everything else, etc. 01:20 hmmmm hmmm last time i checked the build system was pretty decent 01:20 hmmmm but still fragile 01:20 rubenwardy it used outdated CMake rather than a modern approach, with targets and modules 01:21 rubenwardy josiah_wi has being improving that 01:21 hmmmm i didn't know there was an outdated way of using CMake :( 01:21 hmmmm sounds good 01:21 rubenwardy one of the things you can do now is clone Irrlicht-MT to `minetest/lib/irrlicht`, and CMake will pick it up and compile it automagically 01:21 hmmmm ohhh it's a git submodule 01:22 MTDiscord I'm studying CMake build system design to a degree that's kind of crazy. But studying doesn't replace experience... I broke server builds with headers due to a versioning issue I believe and just heard about it. 01:22 rubenwardy it's not, because there's a dislike of submodules 01:22 rubenwardy it also wasn't directly included because the SLoC dwarves MT 01:23 MTDiscord I didn't actually do the lib/irrlicht, NeroBurner did that. 01:23 rubenwardy eventually, it'll probably be swallowed by Minetest - I'm not sure what the plans were for that 01:23 hmmmm josiah if you wanna do experimental stuff with the build system you've gotta have an array of test systems 01:23 hmmmm I used to have a win7 x64 VM, mac osx VM, linux VM, etc. for testing minetest builds 01:23 MTDiscord Yeah, that's my issue. In particular we have to support CMake 3.5 and I'm on 3.18 01:24 hmmmm the mac osx vm was such a pita i think i dropped testing on that 01:24 rubenwardy we use GitHub CI for Windows and Linux VMs 01:24 hmmmm awesome 01:24 MTDiscord I'm thinking about purposefully getting a computer to install an archaic debian on. 01:24 rubenwardy not sure if it supports Mac OS 01:24 hmmmm of course it doesn't 01:24 hmmmm you're not allowed to run mac osx on a VM as per the license 01:25 rubenwardy haha right 01:25 rubenwardy great 01:25 rubenwardy it's like Apple doesn't want people to support their devices 01:25 hmmmm I mean, people have managed to get it running on a VM, but it's fragile 01:26 hmmmm macs are the BMWs of the computer world 01:29 MTDiscord How come we have a MacOSX CI build for Irrlicht now then? 01:30 hmmmm well that's a positive thing 01:33 hmmmm i have no idea how they do it. maybe they set it up so the CI runs on a physical machine? 01:33 MTDiscord No idea. I just saw it today and all I care about is that the light is green lol. 01:46 pgimeno > eventually, it'll probably be swallowed by Minetest - I'm not sure what the plans were for that <-- from my understanding, the plans are to get rid of it, not to swallow it 01:47 pgimeno the long-term plans at least 01:49 pgimeno to that end, the hardest nut to crack is probably going to be the GUI; v-rob said he's willing to take on that. I'm also concerned about the scene graph, but hecks said it's not a problem so... 01:52 kilbith also, the GUI/HUD will be unified under the same API 01:54 MTDiscord >you're not allowed to run mac osx on a VM as per the license 01:55 MTDiscord factually incorrect 01:55 MTDiscord VMs on actual macs are fine 02:19 v-rob "the hardest nut to crack is probably going to be the GUI; v-rob said he's willing to take on that" -- Correction: already taking it on and making good progress too. 02:26 hmmmm >The macOS Virtualbox option is designed for genuine Apple hardware. You will not get community support from Virtualbox if you have trouble with this process, as it’s against Apple ToS. 02:35 v-rob I think I ought to give a progress report: I decided to work on the GUI entirely separate from Minetest so I don't have to worry about any cruft or dependencies that I have to make first (like Irrlicht deficiencies). Currently, I have a simple abstraction over SDL that can be ported to Irrlicht (or OpenGL if hecks gets there first). At this point, I've got a lot of the important basic stuff down and am working on layouting/sizers and the Lua bindings. 02:35 v-rob Once that's all done, I'll work on events and text (I've decided that Unicode support will be as full as possible or bust -- I'm sick of the current situation). Then, make the rest of the elements, a few more features, integrate with Minetest, and destroy formspecs. Nothing too it. 02:35 v-rob too -> to 02:36 v-rob Personally, I want to do a few beta releases on the GUI (via temporary integrations with Minetest in the dev versions or something) so people can mess around with it, find all the bugs, criticize the API, and tell me how terrible it all is. 02:37 v-rob Anyhow, my schedule is rather hectic right now, so don't expect miracles, but it's moving forwards 02:42 hmmmm it seems as though the EULA changed as of OSX 10.8.2: "(iii) to install, use and run up to two (2) additional copies or instances of the Apple Software within virtual operating system environments on each Mac Computer you own or control that is already running the Apple Software, for purposes of: (a) software development; (b) testing during software development; (c) using OS X Server; or (d) personal, non-commercial use." 02:42 hmmmm so apple says that you can have two mac osx VMs, but only on a mac osx host 02:44 hmmmm errrr wait jordach specified "on actual macs" ok nvm :) confirmed then 07:24 nrz Haha that kilbith troll. People dors n'y change in 6 years :p 11:16 sfan5 I'm thinking about purposefully getting a computer to install an archaic debian on. 11:16 sfan5 you run linux don't you? just use a container 11:38 sfan5 i hope the review process got better 11:38 sfan5 eh doubt you'd consider it better 11:39 sfan5 reviews are already slow and people are frustrated with it, tightening it without a better approach wouldn't be good 12:59 MTDiscord I am new to containers, and my dev PC is a total potato. I believe it can run docker, but maybe not an OS inside it. 13:00 MTDiscord I could try it though. 13:01 MTDiscord Docker doesn't run a whole OS, it just runs the applications. It assumes that the inner OS uses the same kernel as the outer one, so it runs only the single shared kernel, but the userland (what differentiates distros generally) can be independent. 13:01 sfan5 with proper containers (not VMs) running a program inside of them is no different than running one outside 13:02 MTDiscord I run tons of docker containers on raspis; it's pretty impressively lightweight. Mostly just costs a little extra disk space. 13:02 sfan5 so it will run even on a potato 15:53 MTDiscord it sounds more like he won't put his money where his mouth is 16:52 MTDiscord Is there a good resource for learning to set up docker containers for this kind of purpose? 16:55 sfan5 I don't use docker actively but I can imagine you can literally pull the debian image and just open a shell inside to do what you want 16:55 sfan5 since the container instances are exist while they're used, not permanently 17:34 MTDiscord I have no idea how to use docker, though. Should I just look up a simple docker tutorial? 17:35 MTDiscord I absolutely do not want to redownload dependencies every time I start the container btw... 17:54 nrz go on docker website, there is plenty of resources on the web 18:30 MTDiscord I cannot find a Linux distribution that has a CMake 3.5 package. De ian Jessie had 3.7. Does anyone know where I can find a distribution with this version? 18:31 MTDiscord It seems weird to have a minimum 3.5 requirement if no distros anywhere have a package for it, so there must be at least one out there. 18:33 sfan5 slackware 14.2 18:33 sfan5 debian 9 comes close with 3.7 18:43 MTDiscord Thank you. 18:47 sfan5 you can build and install cmake 3.5 on a modern distro btw, just extra work 19:26 pgimeno why cmake 3.5? 19:30 MTDiscord it's the lowest cmake version minetest is intended to support 19:33 sfan5 I went through some package version overview of Linux distros and all distros that weren't ancient (cmake 2.8.*) had at 3.5 or newer so I went with that 19:33 sfan5 at least* 19:34 MTDiscord Do we know whether anyone actually uses that version to compile Minetest? 19:34 sfan5 npo 19:34 sfan5 no* 19:35 MTDiscord Do we ever test on it (does it even work)? 19:37 sfan5 no and probably 19:38 MTDiscord Good grief, slackpkg doesn't install dependencies for packages you install. 19:39 sfan5 yes uh slack is quite ... antiquated 19:39 MTDiscord Anyone compiling Minetest on Slackware must be a brave guy. 19:39 MTDiscord If we confirm Minetest doesn't work on Slackware can we raise the minimum CMake version? I'm curious. 19:42 sfan5 if that's a purely theoretical question then the answer is I don't know 19:42 MTDiscord Being the "do everything yourself the hard way" distro is sort of like slack's niche these days 19:43 MTDiscord I just want to check before I continue. Will my work go away if I exit the container? I started it with docker run --interactive --tty bash 19:43 MTDiscord There's no way I'm going to continue this atm until I know my work will be saved. 19:46 MTDiscord I think you didn't use --rm so your work is ... somewhere 19:46 MTDiscord but not necessarily easy to access? 19:47 MTDiscord Usually with docker you script things out, and treat the containers themselves as throwaway, but I believe docker doesn't automatically throw them away for you either 19:47 MTDiscord No idea, I am running in the dark. I've already done more work in the container than I should've. 19:48 MTDiscord A container is an unviable solution unless I can have it save all the packages I need and preferably the file system. 19:49 MTDiscord Docker is supposedly easy, but for me it's a nightmare to work with lol. 19:49 sfan5 treating containers as throwaway is specifically a docker thing 19:49 sfan5 you wouldn't have this systemd-nspawn containers for example 19:49 sfan5 or lxc/lxd 19:50 MTDiscord I just want advice on where to go from here; I'm thinking building CMake 3.5 would be significantly easier... 19:51 sfan5 well I'd say go with systemd-nspawn and get a debian 9 container with cmake 3.7 where you'll do your work in 19:53 MTDiscord With docker, the least throwaway thing is your dockerfile, which is basically the src. Then the compiled image is somewhat more throwaway, and then the running container is most throwaway. 19:57 MTDiscord That was quick. There's a CMake dependency that there's no package for apparently. 19:59 MTDiscord nevermind; it was in another package 21:52 nrz debian 9 will be EOL soon, in 1 year max 22:13 MTDiscord SQLite is a required dependency even if PostgreSQL is enabled? 22:14 Desour it's listed as dependency in the README, so I think so. 22:15 MTDiscord I'm curious why though. 22:16 sfan5 it'd be very inconvenient if there wasn't a minimum standard world format all minetest builds understand 22:16 sfan5 aside from the fact that auth and player database currently only support legacy plaintext files OR sqlite3 (current default) 22:16 MTDiscord No, I'm asking why I have to install SQLite if I'm only using PostgreSQL. 22:17 MTDiscord Ah, got it 22:17 MTDiscord Pretty soon I will have a package list for dependencies on Slackware. :) 22:18 sfan5 https://0x0.st/-toP.mp4 beautiful 22:18 Desour that won't be so beautiful on a remote multiplayer server I guess 22:19 sfan5 you need to live with the latency 22:19 Desour not if we had sscsm 22:20 MTDiscord Isnt there already a CSM for that 22:20 MTDiscord That honestly seems like a job for straight CSM 22:23 sfan5 dunno 22:23 sfan5 I need this to test the server-part anyway 22:39 MTDiscord honestly SSCSM is simple 22:39 MTDiscord you're just making it more difficult than it actually is 22:46 Desour actually, I've recently worked on a branch to implement sending of files with paths and so on. it basically reforms all the media sending stuff (at least the announcing and receiving, I didn't come yet to the data sending) 22:46 Desour the problem is just that there's so much one could improve in this area. eg. that you can use paths for textures 22:46 Desour or using tcp for reliable packets; and implementing some sort of stream packet, to replace the "bunches" in the media sending protocol 22:46 sfan5 sounds like remote_media except no manual setup 22:46 sfan5 (and more complicated) 22:47 Desour something like remote media is one application of stream packets. another would be streamed sounds 22:47 Desour ie. ingame music that doesn't take seconds to load 22:50 Desour and reimplementing the reliable packets with tcp would also bring all the other benefits of tcp. minetest currently doesn't even probe the MTU and just uses packet sizes of 500 or 512 (one is server, the other the client) 22:52 sfan5 plus the downsides of tcp 23:57 MTDiscord At last! I'm building my Docker image, and it'll be able to use a graphical display. I compiled Minetest, but I couldn't test it yet. 23:58 MTDiscord This is still using Slackware. So Minetest does build, with a simple configuration and no sound. 23:59 MTDiscord (I didn't figure out the right vorbis libs yet so I skipped that because it was getting boring.)