Time Nick Message 00:05 MTDiscord i see some sfan5 review comments there that seem to not have been addressed? 12:50 MTDiscord merging #14557 in 5d 12:50 ShadowBot https://github.com/minetest/minetest/issues/14557 -- Add static glTF support by JosiahWI 12:58 sfan5 merging #15068, #15072 in 20m or more 12:58 ShadowBot https://github.com/minetest/minetest/issues/15068 -- [no squash] Rendering and some other fixes by sfan5 12:58 ShadowBot https://github.com/minetest/minetest/issues/15072 -- Menu docs: clarify that images must be escaped to correctly render on Windows by Zughy 13:01 sfan5 oh and also #15046 13:01 ShadowBot https://github.com/minetest/minetest/issues/15046 -- [nosquash] Transparency sort mesh radius fix, and other things by Desour 13:02 Desour you can also click the button for #15045 :) 13:02 ShadowBot https://github.com/minetest/minetest/issues/15045 -- Convert nodedef tests to Catch2 by JosiahWI 13:02 sfan5 sure 13:05 sfan5 let's say 30m then 13:30 rubenwardy Dang, 5d not 5m 13:30 rubenwardy You could just ask sfan5 if he wants to review it or not 13:33 Desour let's hope it's 30 minutes, not 30 months 13:34 MTDiscord 30 millisomething 13:35 MTDiscord 30 metres, duh 13:49 Desour exit 13:52 [MatrxMT] Copyright year is 2022 in README, can anyone push directly to master to fix it? It's silly to make a PR for that 13:57 MTDiscord isn't master a protected branch, so no? 13:57 MTDiscord someone could make a pr and immediately merge it though 14:01 sfan5 we don't have protected branches 14:01 rubenwardy Too cool for protected branches 14:02 rubenwardy We do the opposite, make it hard to merge PRs 🥲 16:08 MTDiscord Can we just increase copyright year indefinitely like that? 16:09 rubenwardy you don't actully need copyright years so can just remove them or link to LICENSE 16:20 Noisytoot Unless there are no copyrightable changes in that year, yes. 16:32 jonadab Note that someone who gets ahold of an older version of the codebase, is not bound by the updated copyright in the newer version. Once the older version is old enough for copyright to expire, that version is in the public domain, even if the newer version is not. 16:33 jonadab So for example just because Steamboat Willie is in the public domain, doesn't mean you can take any Mickey Mouse cartoon Disney has ever made and treat it as public domain. You can't. Just the one(s) that actually entered the public domain. 16:34 jonadab The updated copyright covers the changes. 16:35 MTDiscord I guess git history kinda does that job though 17:27 Krock it might be time to clean up some header dependencies. Modifying IVideoDriver.h recompiles server/* and unittest/* 17:28 sfan5 yeah 17:43 grorp re https://github.com/minetest/minetest/pull/15022#discussion_r1731729621: what do other core devs think? 17:44 grorp I could propose an alternative pure server-side fix based on get_player_window_information, but it would be more code 17:46 Krock protocol bump is fine by me 17:58 grorp I'd like to get #14749 merged to unblock work on other PRs 17:58 ShadowBot https://github.com/minetest/minetest/issues/14749 -- [no squash] Split touch controls from UI by okias 17:59 grorp sfan5: can I still https://irc.minetest.net/minetest-dev/2024-08-18#i_6193573 as an approval? 17:59 grorp +count 18:29 sfan5 yes 18:29 sfan5 merging #15067 in 10m 18:29 ShadowBot https://github.com/minetest/minetest/issues/15067 -- Delete OpenGL ES 1.0 driver by sfan5 18:29 sfan5 (new PR soon that depends on it) 18:33 MTDiscord I think connecting with 2 clients to a server at the same time with local map saving enabled locks the database and can crash one of the clients :^) Dont think that used to happen 18:33 sfan5 highly doubt that's a new bug 18:35 grorp great, merging #14749 in >= 30 min 18:35 ShadowBot https://github.com/minetest/minetest/issues/14749 -- [no squash] Split touch controls from UI by okias 18:36 Krock after many doubts and incorrect indices I can now proudly say that I almost nullified the drawcall overhead for rectangles. I yet need to integrate font rendering for an good first PR. showcase https://i.postimg.cc/8CkmgsZc/grafik.png | branch https://github.com/SmallJoker/minetest/tree/irrlicht_batch_poc 18:37 Krock interestingly, OpenGL could even take care of the clipping (glScissor) 18:39 Krock hmm. does MangoHUD yet not have an option to show up the amount of drawcalls per frame? 18:44 sfan5 grorp: do 15064 too 18:45 sfan5 #15064 18:45 ShadowBot https://github.com/minetest/minetest/issues/15064 -- Main menu formspec escape world name by cx384 18:48 MTDiscord Krock: quite frankly i'm not yet sure that this drawcall overhead ends up mattering since 2d scenery is inherently relatively simple. a formspec would have to be horribly packed to result in terribly many drawcalls. are there good examples where this results in a tangible FPS benefit in a menu? 18:48 sfan5 none of these optimizations matter on a desktop gpu 18:49 MTDiscord note also that the order in which things are drawn matters very much for GUIs so I'm skeptical about anything that might mess with this order 18:49 sfan5 they do matter on low-end mobile gpus 18:49 Krock @luatic CPU usage at 7% normal world, 16% with unified_inventory open 18:50 MTDiscord are you using software rendering? 18:50 Krock OpenGL 4.6 (OpenGL 2.x) on amdgpu 18:50 MTDiscord does CPU usage include GPU usage for you? 18:51 MTDiscord anyways, for any such PR imo it'd have to be demonstrated that it yields an improvement somewhere. i could probably test on a relatively old android phone. 18:52 Krock okay fair. with my changes so far that's only gone down to 14%. corectl reports 0% GPU usage, which is not very helpful. 18:52 Krock *corectrl 18:52 MTDiscord Krock: at a quick glance, how do you plan to ensure that Z index draw order is correctly preserved? for example how do images and rectangles interact? 18:53 Krock @luatic once everything passes through this batch step, they are hopefully in order thanks to the Z index. Although I yet have to test whether that works at all. 18:53 sfan5 you could potential use the z buffer for that but it's not trivial 18:53 sfan5 Krock: i don't think it's that easy. consider overlapping elements 18:54 Krock sfan5: yes I meant to address the Z index problem with the Z buffer 18:54 MTDiscord that seems sensible. make sure you're clearing the scene Z buffer though. 18:54 Krock (where Z index = draw order, basically) 18:55 Krock the scene is at least cleared once by the wield item. possibly once more for the formspec rendering. But I'll have to check. 18:56 Krock disclaimer: I barely have an idea what I'm actually doing :) 19:07 sfan5 i think i need a faster cpu to speed up compiling MT 19:12 Krock I suppose you already disabled LTO? 19:13 sfan5 yea 19:13 sfan5 also using ccache 19:25 Krock https://i.postimg.cc/HsdJYg98/grafik.png fonts do work ... so far. 19:31 grorp merging 19:37 grorp haha, now all of my PRs except one are "rebase needed" 19:41 [MatrxMT] rebase party 19:42 [MatrxMT] do you think we'll be able to ship 5.9.1 this Sunday? Current situation: https://github.com/minetest/minetest/milestone/27 19:43 sfan5 network bugs needs confirmation. I could prepare a PR with the changes at any time 19:43 sfan5 15000 too unclear 19:44 sfan5 14999 not unclear but the cause is obscure, can't fix 19:44 sfan5 15005 could be fixed it doesn't seem so important for 5.9.1 19:44 sfan5 15056 could be duplicate of the network bug (14765) 19:55 sfan5 so to answer your question: if we decide that 80% of the milestone are notabug/cantfix/wontfix, then yes 19:58 rubenwardy It's worth releasing anyway if the remaining issues are obscure. Can always do 5.9.2 later or wait until 5.10 19:58 rubenwardy The issues are there to be investigated, not necessarily fixed in 5.9.1 20:02 MTDiscord Re: 15000, the urgency level has dropped a lot since I first reported it, and I'm now running 5.9 with mostly okay results, no need to roll back to 5.8 now. 20:03 MTDiscord I've been having a hard time finding anyone else to +1 it. I've heard a lot of rumors about 5.9 having "performance issues", but the most concrete corroboration I've found so far was Emoji saying their server went into meltdown ... once. 20:03 MTDiscord It's possible it's just been worked around, but it's also possible that this is just a rare problem that only happens under limited circumstances. I'd still like to see it investigated but it doesn't feel 5.9.1 kinda urgent. 20:14 sfan5 promised PR: #15076 (I estimate 8-12 hours of work) 20:14 ShadowBot https://github.com/minetest/minetest/issues/15076 -- [no squash] Ability to handle vertex and index VBOs separately by sfan5 20:15 sfan5 but please don't expect better performance in average cases 21:11 [MatrxMT] shouldn't we remove .gitlab-ci.yml if it doesn't work anymore nor it's used? Appimages for sure are broken 21:13 [MatrxMT] (ignore the appimage part, that was removed a while ago) 21:30 [MatrxMT] what's core dev consensus about shipping appimages? Or, at least, provide a documentation about how to make one? It's been two days with two people working on it and it's highly frustrating 21:30 [MatrxMT] (no results 🙃) 21:35 ROllerozxa I've been wanting to set things up to produce official appimage builds for minetest for a while now. I have experience with it from setting up appimage builds for principia and mostly know what it takes to bundle up things 21:36 ROllerozxa (not a coredev but just mentioning it) 21:38 [MatrxMT] can I write you privately on Matrix? 21:39 ROllerozxa sure. I assume you know my handle there 21:58 MTDiscord Zughy: I'm doing some builds for AppImages as unofficial for a while: https://github.com/ronoaldo/minetestclient - if that code helps, feel free to use it. If a PR is welcome, I can prepare one to be integrated into the current CI. I personally feel that Flatpak would be the official channel to go for Linux... It will somewhat solve the same issue for potentially more distros. AppImages are a good "portable" version for Linux tho. 22:04 [MatrxMT] ronaldo: o: 22:04 [MatrxMT] yes, appimages are indeed very useful for instance when you can't install anything on PCs (e.g. edu versions for schools) 22:28 MTDiscord I'm on discord, kindly @ me if it is worth opening a PR for that, if the core devs find it useful. I can keep maintaining this particular part too to avoid overwhelming more people. 23:26 [MatrxMT] @ronoaldo check your Mastodon 🙏 23:37 MTDiscord Zughy: checked! thanks