Time Nick Message 00:10 MTDiscord not just firewalls, the big switches that deliver the Internet can fail to do other protocols. and if someone's behind a NAT, then there's no chance they've implemented it for the others. 00:11 MTDiscord in theory The Internet shouldn't care what you stick in IP, and NATs violate every Internet rule there is, but 00:12 MTDiscord Hopefully in IPv6 everything is agnostic but I wouldn't get my hopes up… 00:15 MTDiscord I predict TCP will eventually fall out of use, and UDP with become thee transport protocol, everything building on top. 00:20 celeron55 quic seems to me like a better, more fully featured enet 00:20 celeron55 and enet already is pretty good 00:21 celeron55 of course, a bad implementation can still ruin it. the implementation has to be reasonably compact, portable and modular 00:24 celeron55 i agree about that prediction with TCP. it's too inflexible to be future proof 01:51 MTDiscord u32 type for active objects 12:13 MTDiscord why do you want so many active object ids? 12:15 MTDiscord the engine can't handle that many objects and keep up even a fraction of performance presently 14:13 celeron55 u32 could be (essentially) world-unique 15:19 MTDiscord maybe some forks can? 15:21 MTDiscord and no need to select free ids, always simple increment 15:22 [ what about when it overflows? 15:23 MTDiscord okay, u64 is better 15:44 sfan5 merging #15398 in 10m 15:44 ShadowBot https://github.com/minetest/minetest/issues/15398 -- [no sq] Texture/rendering fixes and improvements by sfan5 18:04 MTDiscord extendable mapblock disk(and net?) format (msgpack or protobuf) 21:26 MTDiscord just wanted to check in real quick: would anyone have a problem with me making CImage ARGB8 only? 21:26 MTDiscord as a compromise I could offer making it a generic CImage 21:28 MTDiscord the upsides of ARGB8 only are of course simpler code / less maintenance, but also getting rid of many switches in possibly hot code paths (e.g. remember the ARGB8-specific optimization we did) 21:28 MTDiscord 16 bit color formats are not something we really care about anways 21:29 MTDiscord so the discussion is really just do we keep RGB8 and i don't see a strong case for that. we probably still want things to be ARGB8 for the CPU so everything is nicely dword-aligned. we could consider sending it as RGB8 to the GPU but that's not much of a saving. 21:41 [MatrxMT] Hi fellow Luanti people, please review #14933 21:41 ShadowBot https://github.com/minetest/minetest/issues/14933 -- Implement an editor to customize the touchscreen controls by grorp 21:44 [MatrxMT] It's the PR with the most hearts of all the PRs currently open btw, in case somebody needs convincing