Time Nick Message 00:04 hmmmm okay 00:04 hmmmm i'll have to take a look 03:46 hmmmm paramat: we can fix the oregen crash by switching to PcgRandom 03:52 paramat ah okay 03:52 paramat good idea 03:54 hmmmm what do you think about 3237 03:55 hmmmm #3237 03:55 ShadowBot https://github.com/minetest/minetest/issues/3237 -- PseudoRandom always generating same sequence with big seed 03:55 paramat hm yeah, a dilemma 03:57 hmmmm oh darn 03:57 hmmmm paramat, I just noticed something about the fractal mapgen 03:57 hmmmm you didn't include mapgen_fractal.cpp in the android makefile 03:58 paramat i did in a following commit 03:58 hmmmm ahh good 03:58 paramat but yes stuff broke for a day 03:59 paramat does 3237 affect core mapgens too? 04:00 hmmmm it affects anything unfortunately 04:00 hmmmm any time you add a new source file, it needs to be added there 04:00 paramat ok 04:00 hmmmm i really do not like the way our android build system operates but it's like that until somebody who is knowledgeable enough about the android build can fix it 04:02 paramat so issue 3237 is only for seeds used in on-gen functions, so just mods and lua mapgens? 04:03 hmmmm the blockseed is 64 bit in all new mapgens i think 04:03 hmmmm it's still 32 bit in v6 04:04 paramat ah yes i remember mgv5/v7 being updated to a new blockseed creation 04:04 paramat anyway for the fractal bug i'll try some different 'getters' and also try using v3fs for some parameters 04:04 hmmmm yeah one of the problems with v6 is that the blockseed used a VERY bad 'hash' of the position coords and seed 04:05 hmmmm the bottom bits had zero enthropy depending on whether the seed was even or odd 04:05 hmmmm ahh okay, getBlockSeed2 is 32 bit as well 04:05 hmmmm it just has improved enthropy 04:11 paramat the fractal mapgen bug repeatedly throws up this number 163148516993831175979008.0 i wonder if that's familiar to anyone .. 04:12 hmmmm paramat, how have you been selecting the fractal mapgen btw? 04:12 hmmmm just wondering 04:13 paramat well when it was just a local branch normally, now of course it's hidden so i use a mod using 'set mapgen params' 04:13 paramat buh was present both ways 04:13 paramat (bug) 04:13 hmmmm oh okay 04:13 hmmmm was about to say, that's pretty telling 04:13 paramat buhhh ... 04:14 paramat getFloatNoEx seems to work in mgv6 04:14 hmmmm first I need to fix the mapgen selection 04:22 hmmmm hmmm this is gonna be hard with the way formspec works 04:23 hmmmm so, my idea was to get the full list of mapgen names as well in create_world_formspec() 04:23 hmmmm if the mapgen name isn't in the publically available list, then check the full list 04:23 hmmmm if it's in the full list, set the contents of the combobox to that mapgen name, but don't add it to the dropdown 04:24 hmmmm maybe even add a label to the side that says "(EXPERIMENTAL)" or something along those lines 04:24 hmmmm however we can't do this without adding a new formspec type 04:24 paramat i noticed i couldn't select a hidden mapgen using .conf, only by a mod 04:25 hmmmm right, obviously this doesn't matter so much for singlenode but for the fractal mapgen this might be a real problem 04:25 hmmmm not really sure how to "fix" this 04:26 paramat currently i like how maogens can be well hidden, and only knowledgeable users know how to select with a mod 04:26 paramat devving a mapgen that is 'in' is a relief 04:27 hmmmm fair enough, i guess 04:27 paramat i have another one i might like to add hidden soon 04:29 hmmmm the way caves are asymmetrical kind of ruins the fractal mapgen imo 04:29 kahrl how about showing the fractal mapgen in the menu, but only if mg_name is set to fractal 04:29 paramat lol, normal v7 tunnels 04:30 hmmmm paramat, at y=62.5 there are abnormal shadows 04:30 hmmmm just dyi 04:30 hmmmm fyi* 04:30 hmmmm how do you replicate this problem? just by joining & leaving the game repeatedly? 04:31 paramat yes either to menu or to OS 04:31 paramat the bug can happen on first session in a new world 04:31 hmmmm and how frequently do you experience the issue? 04:32 paramat it occurs randomly, maybe 25% of sessions 04:33 paramat lighting bugs at y = 63 #2759 in mgv5/v7/fractal 04:33 ShadowBot https://github.com/minetest/minetest/issues/2759 -- Lighting bug at y = 63 in mgv7 and core mapgens in development 04:34 paramat probably due to the alternative 'calc lighting' used in overgenerating mapgens 04:34 hmmmm now is that really the case 04:34 hmmmm the overgenerating mapgens only overgenerate noise 04:34 hmmmm not the actual result 04:34 hmmmm anyway I can't reproduce your problem 04:34 paramat the fractal bug occurs more often in julia mode 04:35 paramat mgfractal_spflags = julia 04:35 hmmmm kudos to making a playable, actually nice looking fractal mapgen 04:35 hmmmm this is miles better than what proller had 04:35 paramat but you'll need to fly east 500-1000 nodes to find it 04:35 hmmmm the most important thing is that it's actually playable 04:36 hmmmm it's not an incoherent jumble of nodes 04:37 paramat the slice param provides a range of different shapes, the julia mode another set of shapes 04:37 paramat all scalable and offsettable for crazy stuff if users want 04:38 paramat anyway i'll try using v3fs for some params in case that helps 04:38 hmmmm eh that's not going to help... 04:38 hmmmm the julia mode seems to be all water 04:39 paramat fly east 500-1000 nodes, if it doesnt appear bug is occuring, re-enter world 04:39 paramat it might then appear 04:40 paramat btw someone else has replicated the bug 04:45 hmmmm yeah sorry paramat, can't reproduce 04:46 paramat no problem, it can take some time to do so 04:48 paramat very frustrating bug *=( 04:49 hmmmm try running with valgrind 04:50 paramat hm i'll have to learn that, about time i did 04:50 paramat now merging 3236 04:54 hmmmm want me to retrofit the decorations to use PcgRandom 04:54 hmmmm ? 04:56 paramat done 04:56 paramat for that ore issue yeah 04:56 paramat oh sorry 04:57 paramat doesn't seem essential, only if you feel like it 04:58 paramat but of course the new PR is better 04:59 paramat i'm up to 100 commits to engine crumbs 05:02 hmmmm heh 05:03 hmmmm hi five 05:03 hmmmm and you originally thought that core development would be too difficult for you to handle 05:06 hmmmm anyway, my initial reaction to the non-user visible mods is to fix that oversight so a mod isn't absolutely necessary to use it, but I think you make a good case for leaving it as-is 05:07 hmmmm non-user visible mapgens i mean 05:08 paramat indeed :} 05:09 paramat as-is is fine for now 05:09 paramat http://i.imgur.com/hFPGIpk.png taste of my rivergen branch 05:11 VanessaE beautiful 05:12 paramat http://i.imgur.com/3Olw5am.png also playing with floatlands in mgv7, easy to add but need to think about how to get up there 05:14 paramat http://i.imgur.com/iNJ0xSP.png for example playing with towers 05:20 hmmmm rivergen is probably the most realistic looking mapgen 05:21 hmmmm i can't help but look at that first screenshot and think, "yeah, that place exists somewhere in colorado." 05:24 VanessaE I thought the same 05:24 VanessaE or more like "wow, that looks rather real." 05:32 VanessaE bbl 05:52 waressearcher2 hmmmm: universe should've used some mapgen anyway 05:52 hmmmm huh? 05:55 waressearcher2 hmmmm: "that place exists somewhere in colorado", that could mean your mapgen very close to that universe used to create itself 06:16 hmmmm alright here's what I think I want to do 06:16 hmmmm first I'll finish off what I'm working on 06:16 hmmmm the lua envlock interface 06:16 hmmmm then i'm going to merge est's LuaSecureRandom 06:17 hmmmm then I'm going to combine all the different Randoms into a single base class, 06:17 hmmmm and expose the same kind of object oriented interface through lua 06:17 hmmmm so that way somebody can do local rand = new Random(type="secure") or type="pcg" etc. 06:18 hmmmm or... 06:18 hmmmm local rand = new SecureRandom 06:18 hmmmm so on 06:19 est31 the first two things sound great, about the last, I'm not sure how you want to unify algorithms with different input sizes/output sizes? 06:20 hmmmm the input/output sizes will be interface defined and abstracted over each implementation 06:21 est31 the specific low level methods should still be available 06:21 hmmmm of course 06:21 hmmmm but in general i think more attention ought to be payed to the interface 06:21 est31 some don't need such, like the secure random 06:22 est31 (methods, not attention :p) 06:22 hmmmm everybody wants to add something to the modding api 06:22 hmmmm they don't give a crap if it's symmetrical or makes sense or is consistent with the rest of it or not 06:23 hmmmm i think we owe it to our modders to give them a consistent, predictable interface 06:23 est31 agreed 06:23 hmmmm like even Win32 api is more consistent than what we have right now 06:23 hmmmm christ.. 06:23 est31 but we shouldnt also hide away algorithm details from them 06:24 est31 at least not here 06:26 est31 we should provide an api for #3237 like expectations "throw this number at it, let it do something with it" as well as one for those who can handle strict specifications of input/output for algorithms 06:26 ShadowBot https://github.com/minetest/minetest/issues/3237 -- PseudoRandom always generating same sequence with big seed 06:27 est31 but yeah, there are horribly ugly examples of what you talk about, like #2658 06:27 ShadowBot https://github.com/minetest/minetest/issues/2658 -- get_look_yaw, set_look_yaw, get_look_pitch, set_look_pitch unable to use same values as input/output 06:28 hmmmm not only that but this is a perfect of example of why people should wait for the client side modding 06:28 est31 btw why did I say "but" 06:28 hmmmm come on man 06:28 est31 there is no but here 06:29 hmmmm there's so much development TODO, such little time :( 06:29 hmmmm I can't do any meaningful amount of development on minetest outside of vacations 06:30 est31 it still should be fun, not an obligation 06:37 est31 what do you think hmmmm, once client side modding arrives, should such callbacks be removed? 06:38 hmmmm which callbacks 06:38 est31 like set_look_pitch for players 06:38 hmmmm that's a callback?? 06:39 est31 API callback or whatever 06:39 hmmmm modding API function 06:39 hmmmm we can't really remove things from the modding api 06:39 waressearcher2 est31: is client side modding in development ? it should decrease lag, right ? 06:39 est31 waressearcher2, it is not, but it is planned 06:40 est31 or dont know 06:40 est31 ask hmmmm 06:40 est31 he had parts of a patch a while back 06:40 hmmmm things that obviously should be client side modding get emulated by broadcasting a mod RPC call 06:40 hmmmm yeah it is in development but I put it on pause for a while to fix bugs 06:41 est31 nrzkt, can you replace the line in server.cpp with 06:41 est31 actionstream<getName() <<" [" << ip_str << "] joins game (id " << player->getPlayerSAO()->getId() << "). " << std::endl; 06:42 est31 that gives you the id, once your wife gets deleted, you can look up the id, and find out if indeed that id was scheduled for deletion 06:47 nrzkt est31: good idea, i will add a else in my condition to handle player 06:48 nrzkt i need to fix my puppetmaster , 200 servers are disconnected at work, and i add this condition :p 06:57 hmmmm ShadowNinja: is there a reason for src/threads.h? 07:25 hmmmm okay, now that I'm actually working with the new threading interface I have a couple of complaints... 18:21 est31 btw paramat congrats to your 100th commit 18:24 Krock \o/