Time Nick Message 02:29 srifqi paramat, we need to know which is 3D noise and which is 2d noise to set the default value. 02:30 srifqi or, maybe we can list which noises use which flags? 02:40 paramat ok 15:55 sfan5 before it's forgotten again, anymore comments/reviews for #6406 ? 15:55 ShadowBot https://github.com/minetest/minetest/issues/6406 -- Make INodeDefManager::getIds return a vector, not a set by sfan5 17:21 Krock sfan5, LGTM 17:22 Krock tested, I supposed? 17:22 Krock -d 17:22 sfan5 it doesn't break spectacularily 17:22 sfan5 yes 17:22 sfan5 what about the wrap_text PR btw 17:23 Krock 1 moment 17:25 Krock well yes. Either needs a notice in the docs or a fix to enforce the limits. Surely, the 2nd way would be more elegant but I'm fine with both solutions 17:25 sfan5 i'll amend the docs then 17:26 Krock also amend the return value "returns a string" away 17:26 Krock leftover from the previous docs 17:29 sfan5 i'll merge both soon then 17:30 Krock ok 17:35 sfan5 also a fix for the linting failure (https://travis-ci.org/minetest/minetest/jobs/274303626#L481) 17:36 Krock will merge #6404 in ~10 mins 17:36 ShadowBot https://github.com/minetest/minetest/issues/6404 -- Network: Remove large parts of deprecated legacy code by SmallJoker 17:37 Krock that's not a style issue. LINT is just being stupid. 17:39 sfan5 i know, the failing travis build was just bugging me 18:10 sfan5 Krock: umm 18:10 sfan5 2017-09-12 20:10:40: ACTION[Server]: User singleplayer tried to log in, but password field was invalid (invalid base64). 18:10 sfan5 2017-09-12 20:10:40: ERROR[Main]: Access denied. Reason: Server authentication failed. This is likely a server error. 18:10 sfan5 singleplayer 18:11 Krock uhm.. 18:11 sofar ouch 18:12 sfan5 doesn't happen on new worlds 18:12 Krock because I'm stupid. old worlds still have old base64 hashes 18:12 Krock working on it 18:27 Krock sfan5, could you please either test https://pastebin.com/raw/99b0hyMU or send a dummy world to check whether it's fixed? 18:28 sfan5 works 18:29 Krock pushing in 5 mins. 19:28 paramat once fixed, can i merge #6408 ? fairly simple mapgen stuff 19:28 ShadowBot https://github.com/minetest/minetest/issues/6408 -- Cavegen: Avoid unsupported biome 'top' or 'filler' nodes by paramat 19:42 paramat i'm now going through old feature requests and marking some 'possible close', please check these at some point 20:30 paramat #6408 fixed and retested 20:30 ShadowBot https://github.com/minetest/minetest/issues/6408 -- Cavegen: Avoid unsupported biome 'top' or 'filler' nodes by paramat 21:26 Megaf I wonder if AMD have a special compiler optimized for its CPUs, like Intel does. 22:01 nerzhul merging #6410 22:01 ShadowBot https://github.com/minetest/minetest/issues/6410 -- Item drop: Tune to land exactly 2 nodes away with level view by paramat 22:02 nerzhul sfan5, sofar i updated #6351 (got some time), now we have a ModChannel object in lua 22:02 ShadowBot https://github.com/minetest/minetest/issues/6351 -- Implement mod communication channels by nerzhul 22:03 nerzhul oh i need to rebase it seems 22:06 sofar Megaf: afaik, no, they don't (llvm and gcc are cool enough, anyway) 22:15 Megaf is ShadowNinja still alive? 22:16 Megaf #3699 could be rebased and merged IMHO 22:16 ShadowBot https://github.com/minetest/minetest/issues/3699 -- Use SQLite's R*Tree to store rollback node positions by ShadowNinja 22:16 Megaf I may be able to test that if rebased 22:16 rubenwardy are there any benchmarks of the backend 22:17 rubenwardy intuitively, a key-value store would be more suited as it's supposedly more optimised for key look ups 22:17 Megaf paramat, good job revising old issues. Maybe we get some more stuff done now. 22:17 paramat SN's not active in MT currently 22:17 Megaf too bad 22:18 paramat some of his old PRs are in a bad state and might be closed, as in, so much rebasing might as well start again 22:20 nerzhul KV store are more optimized if key are text generally 22:20 paramat (well, i've marked some for close in a month unless there's a reason to keep them open) 22:20 nerzhul relational DB B-Tree index are more performant on integers than hash in KV store 22:20 rubenwardy interesting 22:20 rubenwardy would be good to have stats though 22:21 rubenwardy same map, convert to all types, then see how long it takes to load all the blocks 22:21 rubenwardy then maybe try a random order for loading 22:21 Megaf is rollback using sqlite in minetest? 22:21 rubenwardy hmmm, not sure how I'd program that 22:21 nerzhul it's difficult as disk cache and various backends are using various algorithm and it depend on what is currently in memory and disk cache, to do a real test you should restart the server for each test 22:22 nerzhul random order is not very important, in B-Tree and hash keys it's relatively easy to find them correctly in indices 22:23 nerzhul rubenwardy, i hope the object in #6351 is good it owns the mod channel things (except creation which generate the ref lua object ) 22:23 ShadowBot https://github.com/minetest/minetest/issues/6351 -- Implement mod communication channels by nerzhul 22:23 nerzhul this PR is very huge and feature should be frozen to permit a merge and have little pr to add mor efeature after, else we will never see it in git tree :p 23:53 sofar ShadowNinja: mc2mt still segfaults for me when I build it on Linux (aside missing DIR_DELIM somehow). Can I bug you with some segfaults later?