Time Nick Message 14:05 sfan5 merging #17405 soon 14:05 ShadowBot https://github.com/luanti-org/luanti/issues/17405 -- Update Gradle and fix Android build scripts by sfan5 14:45 sfan5 https://static.wikia.nocookie.net/machikado-mazoku/images/8/82/Metako.jpg/revision/latest 14:59 MTDiscord Oh no. Compatibility-breaking update for Android users app builders o_o 15:27 sfan5 also pushing https://github.com/luanti-org/luanti/commit/7f2939177993089cf45a20b62967798c8e6fb3aa once build passes 16:20 Krock When's a good time for the locale update + RC build? 16:21 Krock FYI bugfix https://github.com/luanti-org/luanti/pull/17348 is still open 16:22 Krock unless someone would like to make a counter-proposal 16:54 sfan5 well we need to decide if we are going to have the visual effects PR or not, before proceeding with a weblate update 17:02 Krock technically we're in feature freeze... 17:02 Krock though I'd rebase and address the comments now if we really want to have this 17:02 sfan5 thats the thing to decide I mean 17:04 Krock https://strawpoll.com/bVg8BEGG1yY 17:05 Krock giving it 2 hours to decide. will rebase the PR in the meantime 17:07 sfan5 shouldn't this vote rather be inside our org discussion? 17:10 Krock yes, you're right... https://github.com/orgs/luanti-org/discussions/170 17:12 sfan5 maybe we can ping rubenwardy, nore, @luatic, @Zughy, @cx384 to vote on this too 17:12 [MatrxMT] already voted 17:13 Krock dear god. That PR is much older than the depth of my git clone 17:13 sfan5 I hope we don't end in a tie 17:14 rubenwardy It can be merged in but the freeze period should reset 17:14 [MatrxMT] I was scared by the "there might be bugs" but to be honest the celestial vault was quite broken for a long time until I started fiddle with it. So I suppose that worst case scenario bugfixes will follow? 17:33 Krock the worst-case bug scenario is that some Lua API does not work due to incorrect push/pop use, or shader incompatibility on some devices or combinations of settings 19:00 MTDiscord i have some pending comments but i think none of them was critical, let me check again 19:01 MTDiscord well except for the thing sfan / krock already pointed out 19:01 MTDiscord the lua -> cpp conversion also is more lax than i would like 19:03 Krock The poll has 5 voltes (unchanged for at least an hour). Shall we give more time or go ahead as the majority decided? 19:03 Krock for the sake of simplicity .. let's just continue without wasting more time. 19:04 MTDiscord just voted for yes, without knowing that this would result in a perfect 3-way tie 🙃 19:04 Krock > I hope we don't end in a tie 19:05 Krock perfect so let's wait for a 7th vote as tie-breaker 19:05 sfan5 we could also just wait until tomorrow 19:38 MTDiscord <.zenonseth> I'm somewhat invested in the visual effects vol.2 partly because it adds artificial_light color setting - but sunlight/moonlight color and more importantly, no reset flag, and no day-night blending of it, (also no smoothing/damping) - all things that I do + moon/day color in https://github.com/luanti-org/luanti/pull/17314 19:39 MTDiscord <.zenonseth> so I'm curious to see if its merged since i can rebase my changes on top, though one important distinction is that i was using hex colors, but can re-do them 19:41 MTDiscord <.zenonseth> like it does a lot, but the ariticial_light api it has is less flexible than what i had imho 19:41 MTDiscord pretty much the only reason to use rgb tables instead of ColorSpec should be that it needs more than the 0-1 range with 256 steps 19:41 MTDiscord if the API is marked experimental, it should be no problem to extend it later, even in a backwards-incompatible way 19:43 Krock it might've gone faster if that PR didn't combine multiple features in one 19:44 Krock s/gone/went/ 19:46 MTDiscord <.zenonseth> +1, my general work experience too (would have just reacted 👍 but that doesn't carry over to irc/matrix) 19:50 sfan5 honestly the whole PR would be merged already if it wasn't "let's put ten features into one PR" 19:50 sfan5 s/whole PR/entire features/ (to be pedantic) 20:06 MTDiscord while fair, does 10 features in one pr take longer than 10 prs with one feature each given how long some prs can take in luanti 20:07 MTDiscord if you have 10, i assume at least one would find luantis curse of prs sitting for years. so kinda doubting the projection 20:12 MTDiscord if we do some quick polling of the pr list https://github.com/luanti-org/luanti/pulls/SmallJoker, https://github.com/luanti-org/luanti/pulls/cx384, https://github.com/luanti-org/luanti/pulls/sfence. all prs that require one rather than two approvals, this would indicate no 20:14 Krock I'd say that 10 separate PRs take longer to merge than 1 large PR. However, having 70% smaller ones merged will be way faster than 100% of the large PR. 20:14 sfan5 okay then let's say the average "time to merge" of 10 PRs will be much smaller than 1 big one 20:14 Krock ^ that 20:16 MTDiscord <.zenonseth> i mean also to point out that smaller, focused PRs are less prone to errors slipping in as they require less reviewer context switching - its a thing every programming job i've had in past 10+ years has agreed on 20:16 Krock partially relevant, when it comes to topic-bound merge times: https://raw.githubusercontent.com/rollerozxa/mt-wiki-dumping-ground/refs/heads/master/devwiki/Merge_Chance_Complexity_Diagram.png 20:20 MTDiscord So my toggle for rendering the game upside down is not gonna happen? 20:20 sfan5 only if you add an API 20:20 MTDiscord I like how android gets it's own section 😆 20:21 Krock imagine macOS above Android. it scaled by the amount of people working on it. 20:21 MTDiscord Well am I allowed to update the chart? 20:22 MTDiscord if the ios pr ever gets merged in the next decade, might be above macos 20:22 Krock it's an archived image of the dev wiki. I license it under CC 0 so do what you want. 20:23 Krock I also created this .. if it happens to be useful to anyone: https://github.com/rollerozxa/mt-wiki-dumping-ground/blob/master/devwiki/player_object_look_horizontal_vertical_yaw.png 20:23 MTDiscord including hud? 20:28 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/1535384069305540791/image.png?ex=6a779158&is=6a763fd8&hm=f07e8a68fc6b05f13bee8097f140048cf6ec58284063f03d7c6a70d57f252039& 20:29 MTDiscord Is Vulkan this hard to support ? Can't SDL2 use Vulkan ?... 20:30 MTDiscord You have to essentially bolt it on 20:31 sfan5 you spend a lot of time rewriting a lot of things and in the end it's not any faster 20:31 sfan5 or: it *is* faster in the end but that's only because you were forced to improve the opengl code on the side (which you could have done directly) 20:33 MTDiscord So Vulkan is just what OpenGL could have been ? I request a merge of these standards >:( 20:33 MTDiscord runelite's gpu plugin is using opengl 4.3+. And they somehow got rendering threads to work with that 20:34 sfan5 simply put, Luanti is not at the point where there are realistic performance goals that can't be achieved with OpenGL 20:35 MTDiscord and where would larger worlds go 20:35 MTDiscord Exactly, that's why it off the difficulty section 20:35 MTDiscord Well larger worlds are already done so they don't go on that chart 20:36 MTDiscord iirc isnt it easier/better to do vulkan on macos via moltenvk vs opengl via angle? 20:36 MTDiscord for greenfield 20:36 MTDiscord It certainly is 20:36 MTDiscord may very well be wrong on that 20:38 sfan5 dunno about the state on macos 20:43 MTDiscord Why is 17314 10 different features? It looks like it's only changing 3 values in the shader