Time Nick Message 00:26 * Megaf is listening 00:26 * Megaf is impressed, the dmg for mac is just 9.4 MB 00:26 Megaf (of voxelands) 00:50 Megaf celeron55: Quite interesting podcast indeed 01:27 paramat hmmmm, new pull request for mgv5 overgen, so far seems to work okay https://github.com/minetest/minetest/pull/1842 01:29 paramat also, when i access eased noise at point i get a warning during build, pasted here https://github.com/minetest/minetest/pull/1834#issuecomment-63128703 02:15 hmmmm uh oh, that can't be too good 02:16 hmmmm lol 02:17 hmmmm paramat, NoisePerlin3DEased() always returns 1 02:17 hmmmm that's because I messed up a parenthesis in the macro and the resulting expression happened to be valid since the last parameter is optonal 02:17 hmmmm so it's float noise = (f(x,y,z), 1) 02:20 hmmmm try it now 02:20 paramat okay good! 02:20 hmmmm hah that was quite a gotcha 02:24 paramat *looks* ahh i was actually looking at those brackets wondering if they were correct 03:08 paramat hmmmm, it works now. the eased top-down per-node search seems fast, timed earlier at ~1ms 04:27 Zeno` I love your feedback on #1819 hmmmm 04:27 ShadowBot https://github.com/minetest/minetest/issues/1819 -- Latest HEAD (18c583e) crashes when using get_perlin_map with an octave greater than 3 and spread of 1. 04:42 hmmmm who is responsible for the redis db? 04:42 hmmmm Database_Redis::Database_Redis() is suspect 06:43 hmmmm lol 06:43 hmmmm implementing an .include directive to config files is a lot harder than i first imagined thanks to the updating 06:43 hmmmm it's done in such an asinine, cockamamie fashion 06:45 hmmmm I GET what celeron is trying to do here, but it's not worth it and it adds a possible TOCTOU bug 07:03 sfan5 hmmmm: me 07:28 hmmmm nevermind, i thought there was a possible bug from not checking the return value of readConfigFile 07:28 hmmmm although the port number could probably be made into a #define constant rather than just a numerical literal 07:28 hmmmm and address could probably default to localhost 07:29 sfan5 why would anybody want to change the default port? 07:31 hmmmm it has more to do with clarity than anything else 11:17 Zeno` can anyone look at the solution provided in #1839 please? 11:17 ShadowBot https://github.com/minetest/minetest/issues/1839 -- Crash on shutdown. 12:32 Zeno` Ok I am merging it 14:52 celeron55 lol, is 1839 the thing that was discussed about for quite long here? 14:52 celeron55 i was going to suggest a missing virtual destructor but then i thought people had surely checked those 8) 15:02 Krock #1839 15:02 ShadowBot https://github.com/minetest/minetest/issues/1839 -- Crash on shutdown. 15:03 Krock yeah. solved now 17:48 Sokomine hi. i still have a problem that can't really be solved well, and i'm a bit cueless as to how to proceed (apart from ignoring the problem and claiming it can't be done reliably): i want to mirror buildings along the x- or z-axis in order to increase variety. as long as full nodes are concerned, that's no problem at all. but after that, it gets complicated and depends on the actual shape of the nodebox/model. there may not even e 17:53 Sokomine just got told that something was missing, so: there may not even exist a node that might act as a mirrored representation 17:54 Sokomine as far as doors go, there is a mirrored representation - which is another node 18:18 kahrl Sokomine: a low-tech solution is to simply have two versions of each schematic - a normal one and a mirrored one 18:19 kahrl that might even be best in terms of performance, as you don't need to map node ids at runtime 18:22 Sokomine that might be best, yes. it wouldn't require any additional ressources. still - some nodes do not require a mirrored version, and there'd be no helpful naming convention 18:24 Sokomine it doesn't sound complex at first to mirror a house. it's just that it ends up beeing exception after exception (and then there are mods/mapgens/games which call the same things diffrently for no apparent reason..) 19:54 selat Hello, what kind of benchmarks should I provide to show the difference in speed between two implementations of pathfinder? 19:59 VanessaE I would think the amount of time to get from point A to point B, the number of nodes crossed, maybe the amount of CPU used during that time, and the highest dtime recorded? 21:02 paramat hi nore, sfan5, this pull needs to be merged https://github.com/minetest/minetest_game/pull/339 the issue is causing significant slowdown in mgv5, mgv7 and all lua mapgens, and creates unwanted flora whenever default nodes are added with the biome API 21:03 rubenwardy How long before mgv7 is considered finished? 21:04 VanessaE rubenwardy: about 5 years :P 21:06 rubenwardy Lol 21:07 rubenwardy Better than when Minetest is considered finished 21:07 rubenwardy The estimate is asymptotic as time goes on. 21:07 PenguinDad rubenwardy: software is never finished 21:08 RealBadAngel indeed 21:08 Krock wait. pong and tetris aren't finished? 21:09 PenguinDad Krock: these are only discontinued 21:09 Krock oh! 21:10 PenguinDad The only software which might be finished is Hello World :D 21:10 rubenwardy Ha. There is always something you can add to it, and then it ends up like gmacs. 21:11 Krock you can not extend the hello world program. it would go over its target 21:11 sfan5 paramat: I already gave my ok 21:12 paramat yes thanks for that 21:13 PenguinDad Krock: you could make a structured virtualized extended real-time hello world :P 21:14 * Krock puts PenguinDad in the trash 21:14 * Krock rinses it 21:14 Krock flush flush 21:16 paramat Blockmen has been elusive recently, so i'm relying on you and nore 21:17 nore paramat: I'm ok with it too 21:18 kaeza PenguinDad, http://www.gnu.org/software/hello/ 21:18 paramat thanks! 21:40 hmmmm hrmmm! 21:40 hmmmm I wonder how that guy figured out what the issue was for the crash on shutdown 21:41 hmmmm there was no compiler warning about it 21:47 hmmmm lol what the hell https://github.com/freeminer/freeminer/commit/14f4842fdb655787d0e87bd302b42b20ab5d57e4 21:49 VanessaE what's that do? 21:49 hmmmm creates a memory leak in an attempt to work around that odd double-free-on-shutdown crash 21:49 VanessaE aside from prove how childish his commit messages can be :P 21:49 hmmmm or as proller would say, 21:49 hmmmm "Fixed in freeminer ;)" 21:50 VanessaE haha 21:50 sfan5 ಠ_ಠ21:50 sfan5 what 21:51 celeron55 that's freeminer in a nutshell 21:53 celeron55 i'm actually laughing right now, love it 21:54 hmmmm I have a decent way of sabotaging freeminer (at least on non-freeminer servers) 21:54 hmmmm but it requires downloading and running binary blobs from a server 21:55 celeron55 we don't need to sabotage freeminer, it sabotages itself 21:55 hmmmm actually we could make this open source 21:56 hmmmm hrmm 21:56 hmmmm or not 21:56 celeron55 freeminer.png: http://i.imgur.com/BRQwc84.png 21:56 celeron55 i'm sure i will use this some day 21:56 Krock celeron55, nice sentence. may I put it into my forum signature? :D 21:56 hmmmm any sort of signing mechanism would trivially work 21:57 sfan5 celeron55: do you have different font settings, looks different for me: https://cdn.mediacru.sh/pNYFQkWw5bFS.png 21:57 Krock ( he reverted the cmmit afterwards ) 21:57 hmmmm I was thinking of minetest downloading a module downloaded from the servers that periodically scans for strings in the working executable's memory and if it sees freeminer, changes it to "screwed up miner" or something 21:57 celeron55 and competition is for the best anyway; if someone can compete with minetest by adding memory leaks, then maybe minetest isn't very competitive... 21:57 sfan5 s/,/?/ 21:58 hmmmm and it sends back a signed hash of what's being executed 21:59 Krock hmmmm, nice idea 21:59 hmmmm can't block execution without modifying the executable in memory, which means changing the signature 21:59 hmmmm so this way it wouldn't be dependent on the minetest executable's signature, just the module loaded 22:00 hmmmm hmm 22:00 hmmmm freeminer could work around this messily by allocating all strings that say "freeminer" in a manually-allocated piece of memory with read-only permissions, and the segfault handler would just detect the change and it'd modify the text back to freeminer 22:01 RealBadAngel http://i.imgur.com/5cQT16m.png 22:01 RealBadAngel how do you like that? 22:01 sfan5 why would it segfault on read-only memory? 22:01 hmmmm we could hook the segfault handler first 22:01 Krock RealBadAngel, after the wieldhand, the lua entities? 22:01 hmmmm writing to read only memory is a memory violation 22:02 sfan5 oh you want to write somewhere 22:02 hmmmm kool 22:02 sfan5 makes sense then 22:02 hmmmm RealBadAngel, nice 22:02 RealBadAngel this is slightly fine tuned wielded 22:02 hmmmm yeah, sfan5, we're overwriting strings containing the phrase "freeminer" to "f'd up miner" or something 22:02 hmmmm in this hypothetical protection system 22:03 sfan5 if you want it to be secure you need to get TPM involved 22:03 hmmmm of course 22:03 hmmmm there's lots of ways this could be made bulletproof, but all of those require cryptographic signing of the main executable which is not plausible 22:04 hmmmm and of course all of it is extremely platform specific 22:04 celeron55 so, how about something that actually benefits players instead? 8) 22:04 Krock ^^^ 22:04 hmmmm sounds good to me 22:07 * rubenwardy hugs RealBadAngel 22:07 rubenwardy Small steps to improving graphics. 22:09 RealBadAngel makin a vid right now 22:26 RealBadAngel https://www.youtube.com/watch?v=atJ_lovDYWs 22:27 RealBadAngel theres small glitch visible on the sword, extrusion code needs to be tuned a bit 22:27 RealBadAngel but imho that looks GREAT 22:36 rubenwardy Add setting to customise stack max - https://github.com/minetest/minetest/pull/1843 23:08 paramat getGroundLevelAtPoint: 11us :]