Time Nick Message 09:16 nrz_ sfan5: new callbacks and new Lua state specific for emerge callbacks ? ? 10:37 sfan5 the latter 11:06 rubenwardy fuck github for the "force pushing resets updated_at" thing 11:06 rubenwardy it's so annoying 11:30 erle what exactly is annoying about it? 11:31 rubenwardy it breaks sorting in the PR list 11:31 erle well, in my experience, a force push is interesting from a “should i look at this again” POV (which is why i sort by updated) 11:31 rubenwardy a force push _to master_ resets the last updated of all PRs 11:32 erle oof 11:32 erle wait why would anyone ever 11:32 erle but yeah 11:32 rubenwardy wouldn't be so bad if they had statuc_changed_at or something (where status = open / closed / merged) 11:32 rubenwardy so I can sort PRs by merge/close date 11:33 erle indeed 11:33 rubenwardy also last_changed or something for the patch 12:42 kilbith well shit github is translated now 12:55 kilbith https://i.imgur.com/rvO0eV7.png 12:55 kilbith need to finish this 13:40 erle kilbith looks nice. odes it keeps the area forceloaded? 13:41 erle kilbith also could this be used like a HUD element? think “back mirror for car” 13:41 erle like the one where you view the direction you are coming from 13:49 kilbith I still have to work on the area forceload (Emerge?) 13:49 kilbith HUD element, yes that's the plan 13:50 kilbith but I'm thinking at an abstract camera class whose GUI and HUD inherit from 13:51 erle kilbith, look at minetest.forceload_block() and minetest.forceload_free_block() 13:51 kilbith I only want the client to be involved here 13:52 kilbith in any case I'll need some cooperation 13:54 kilbith I also wonder if camera position == world position 13:54 erle i doubt you will be able to do a security camera mod well if the client needs to be in range 13:54 kilbith yes true 14:20 sfan5 why not delegate the responsibility to load the relevant blocks to mods that'd be using this feature? 14:24 erle sfan5, can't do that if kilbith is merely providing additional cameras on the client, can you? 14:24 erle to me it looks like a variant of the freecam thing that fleckenstein (?) made for dragonfire 14:25 erle if you have it active you can move the camera freely among the map parts loaded on the client 14:26 rubenwardy This looks like a scripting feature? Like, you'd use it for a CCTV node or smth 14:26 rubenwardy It's not just force load but also client loaded 14:26 erle anyways, it begs the question: how does a client get a map update for something outside of its view range? can the server just send that? 14:30 sfan5 erle: last time I checked security cameras don't fly 14:31 sfan5 in other words I don't expect the client to have direct control over where those additional cameras point 14:32 sfan5 well perhaps look direction, but not movement 14:33 kilbith why not delegate the responsibility to load the relevant blocks to mods that'd be using this feature? 14:33 kilbith I've considered that 14:33 rubenwardy Anyway, it makes sense to do this in multiple stages. It's fine for the first version to require the client to be nearby the camera, and then later versions to support loading areas etc 14:35 sfan5 on second thought multiple cameras totally breaks the "the client needs to know only about blocks/objects/changes near it" assumption so cooperation of the server is absolutely needed 14:36 rubenwardy Yeah 14:36 erle sfan5 security cams don't fly. i just wanted to point out that the effect of freecam and what kilbith builds is the same in the end: you can use it to look behind walls or so, because the client actually knows what is there, but remote viewing is impossible. 14:36 rubenwardy Also, was wondering how occlusion culling is done - could be issues if it's scene graph based 14:37 erle stuff like ”inside the box” could very well use that, the boxes are small enough, after all. 14:37 sfan5 transparency sorting won't work, at least 14:38 sfan5 (you'd have to do it once for each camera position) 14:38 erle that's far from “won't work”, given that the workload increases linearly only 14:39 erle if i am not mistaken, every client-side approach to additional cameras imposes similar penalties 14:52 kilbith also thinking at `player:set_active_camera(cam)` to create cutscenes 14:57 kilbith drawing on node is also entirely possible as I can set a RenderTarget to a ITexture 14:58 kilbith so we could entirely imagine a scene drawn on a screen node 15:00 erle mirror mirror on the wall, how would such a camera texture appear in a cam scene after all? 15:04 rubenwardy users are getting irr::core::list compile errors when compiling 5.5.1, there needs to be a fixed version for irrlicht in the README.md 15:16 Pexin "security cameras" don't fly, but drones do. how about a remote-controlled robot mod the player sends into a radiation zone to make it safe to re-enter? 15:44 sfan5 rubenwardy: I suppose people don't read the release notes 15:45 sfan5 even better cmake should detect compatibility and error out 19:19 sfan5 merging #12414, #12396 in 10m 19:19 ShadowBot https://github.com/minetest/minetest/issues/12414 -- Mapblock Mesh BspTree: Increase the depth of block-level splits by x2048 19:19 ShadowBot https://github.com/minetest/minetest/issues/12396 -- Sanitize player position and speed server-side by sfan5 20:53 sfan5 the gitlab CI sure is slow, 25 minutes for something that takes 8 minutes on github 20:54 rubenwardy I don't see any caching 20:54 sfan5 neither does github 20:54 rubenwardy ah 20:55 sfan5 pushing a commit to fix the pipelines, just -4 +4 lines 20:57 rubenwardy at my old job, we used Azure Pipelines. Something much simpler than what we do here took 35-50 minutes 20:58 rubenwardy well, it was also running on a windows worker so I suppose that checks out 22:02 Zughy[m] this "completed/not planned" issue difference on GitHub is so annoying and imho not necessary 22:03 Zughy[m] Some core dev put a "Possible close" on an issue and, instead of just closing it after 30 days, I'm supposed to understand whether is "Possible close" because it was completed or because it's not planned. And if I make a mistake, then some user points that out to me (Iike it's already happened in these days) 22:04 Zughy[m] Like if I didn't have enough work to do already 22:16 erle i maintain that “possible close” without a comment is not helpful because it is used for widly different purposes, like “does this bug still exist?” or “is this even important?” or “does anyone understand this?” 22:55 paradust fyi irc/discord bridge has been down for a while now 23:00 kilbith good thing 23:00 kilbith it has gotten much less noisy 23:05 Zughy[m] should I close #10221 as not planned or does any core dev want to remove the "Possible close" tag? I'm uncertain considering rubenwardy comment 23:05 ShadowBot https://github.com/minetest/minetest/issues/10221 -- Increase get_gametime precision 23:10 erle you added “possible close” yourself, so you should know what you meant with it ig 23:11 Pexin sub-second precision timers persisting between game loads seems.. inherently unreliable for any real world case 23:13 Pexin maybe I'm misunderstanding "persistence" there 23:33 erle uh what 23:34 erle Pexin imagine exiting the game only to load it back up