Time Nick Message 00:33 VanessaE ok, figured it out. Disregard that other pull request. Here's the corrected one: 00:33 VanessaE https://github.com/celeron55/minetest_game/pull/81 00:34 VanessaE this does everything anyone should reasonably want from mese without breaking the new paradigm. 01:09 VanessaE forget #81. I did it wrong. 01:09 VanessaE ignore it. delete it. throw it in the bin. 01:09 VanessaE use this instead: 01:09 VanessaE https://github.com/celeron55/minetest_game/pull/82 03:36 Exio this 'fix' https://github.com/celeron55/minetest/commit/40dac4cde4782943eb00f7bfc7ce2d99135be7c9 makes my minetest to fail to compile with that : http://cadoth.net/~exio4/mt/make.log 03:36 Exio reverted it locally == everything working 03:37 hmmmm ooh, clang? 03:38 Exio i used clang for making it detailed 03:38 Exio http://cadoth.net/~exio4/mt/make.gcc.log 03:38 Exio with gcc, the same error with less info 03:38 hmmmm so what version of irrlicht are you using 03:39 Exio sec 03:39 hmmmm so you don't even know!?!? 03:39 Exio i did't checked 03:39 Exio s/ed// 03:41 Exio 1.8.0 03:41 Exio hmmmm: ^ 03:42 hmmmm so i guess irrTypes.h didn't have u64 defined! 03:43 hmmmm can you manually verify that? 03:43 Exio how? 03:43 Exio oh, wait 03:43 hmmmm grep u64 /usr/include/irrlicht/irrTypes.h 03:43 Exio yep, i don't have u64 defined 03:46 VanessaE neither do I, but it built okay here - or at least, I'm a few commits past that already 03:46 VanessaE (irrlicht 1.7) 03:46 Exio i think the problem is the < 03:46 Exio 8<8 == false 03:47 VanessaE that would tend to be accurate :) 03:47 VanessaE try it with a <= ? 03:49 Exio and worked 03:49 Exio :P 03:50 hmmmm well of course that'd work 03:50 hmmmm that's no solution though 03:50 hmmmm i'm not exactly sure what version of irrlicht it had been introduced; here's their original patch: http://sourceforge.net/tracker/?func=detail&aid=3493502&group_id=74339&atid=540678 03:51 VanessaE wouldn't it make more sense to check for the presence of this definition directly and *then* define it if it doesn't exist? 03:51 hmmmm obviously. 03:51 hmmmm we would do that if we could. 03:52 VanessaE what's the rule on defining a variable and deliberately overflowing it? 03:52 VanessaE what would the compiler do then? 03:52 Exio 1.8.somenumbernotzero :P 03:52 hmmmm gotta find their changelog 03:53 hmmmm vanessae, integer overflow behavior is defined (for unsigned types) as being equal to the result modulo (MAX_UINT + 1) 03:54 hmmmm it is undefined for signed types to my knowledge 03:54 VanessaE hm 03:55 VanessaE I gotta learn more C++ soon. 03:57 hmmmm the irrlicht changelog doesn't say anything 04:00 VanessaE only way I can think to test for such a thing would be convoluted and perhaps prone to error. 04:05 hmmmm it can't be done on runtime 04:05 hmmmm if that's what you're saying 04:05 hmmmm i'm already trying to figure something out 04:06 VanessaE damn. 04:06 VanessaE well there's always grepping through the source as you did, in some sort of preprocessor step :) 04:07 hmmmm that could be possibly done, i hardly know how to work with cmake though 04:07 hmmmm it'd be part of the configure script 04:08 VanessaE life was easier when all you had was string, 16-bit int, and float. :-) 04:08 hmmmm if it were up to me, i'd personally use stdint.h types and that's all 04:08 hmmmm no dependence on irrlicht bull 04:08 hmmmm it's obviously too late for that :/ 04:08 VanessaE yeah 04:18 hmmmm alright, so this is the amount of configurability you're going to be looking at through lua with the V6 mapgen: http://pastebin.com/v98qrFC0 04:19 hmmmm you'll be able to modify the offset, scale, spread factor, seed offset, number of octaves, persistance factor of all the uses of perlin noise, you'll be able to choose which features are used on the map, the frequency/size at which a desert occurs, the water level, and the chunk size 04:20 hmmmm although I am thinking i ought not to let chunk size be user modifiable 04:20 VanessaE agreed on that last bit. 04:20 VanessaE that could seriously break things 04:21 VanessaE especially if the user wants to share his or her map. 04:21 hmmmm well actually if any of these variables are present within map_meta.txt, it'll take precedence over what's set in Lua 04:22 VanessaE good idea. 04:22 VanessaE then if the user goes fiddling with the settings, it is they who broke it and not some random mod. 04:23 VanessaE the number of adjustable params looks good though 04:23 hmmmm right - and that's only for the V6 mapgen 04:23 hmmmm V7 puts this amount of configurability to shame 04:24 VanessaE heh 05:39 VanessaE celeron55, hmmmm: that emergethread "fix" was ineffective. 05:39 VanessaE server's occasionally getting stuck where it can't load the map for anyone 05:40 VanessaE but chat is real time, and the server's CPU usage is so-so (high 30's to low 40's). 05:41 hmmmm it's not the same problem then 05:41 VanessaE so two independent problems then? 05:42 hmmmm it must be. 05:43 VanessaE client isn't receiving hardly anything packet-wise - one every several seconds, unless I pan around (then it jumps to tens per second) or someone moves. 05:53 VanessaE map loaded a bit... 05:53 VanessaE server CPU dropped by about 10% at the same time. 06:02 VanessaE I've got it running under gdb now.. maybe I can relay something the next time it gets stuck 06:25 hmmmm the next time it gets stuck you do ctrl+c, 'thread apply all backtrace', then continue really quick 06:25 hmmmm and you pastebin what it prints out 06:25 VanessaE sure 06:25 hmmmm are you using a debug version? 06:25 hmmmm 'cause it'll be useless otherwise 06:26 VanessaE I'm not sure if this is a debug build or not (with my luck, it is not) 06:26 hmmmm well you can easily tell 06:26 hmmmm when you first type 'gdb minetestserver' on the last line of output it'll tell you if it was able to find symbols or not 06:26 hmmmm if it found symbols, then it's a debug build. 06:27 VanessaE Reading symbols from /usr/local/bin/minetestserver...(no debugging symbols found)...done. 06:27 VanessaE damn. 06:28 VanessaE -DCMAKE_BUILD_TYPE=Debug ? 06:30 hmmmm yeah 06:30 VanessaE ok, getting into place now. 06:31 VanessaE Reading symbols from /home/vanessa/Minetest-Textures-and-mods/minetest/bin/minetestserver...done. 06:31 VanessaE ok. 06:31 VanessaE run. 08:38 celeron55 also, it is probably useful to continue it a few times and print a couple of backtraces 08:38 celeron55 or, more like possibly 13:33 Jeija https://github.com/celeron55/minetest/pull/366 << Opened a pull request for a list of servers in the Multiplayer tab 14:11 thexyz -1 for mesecons.net, and i also dislike this idea in general 14:18 doserj the local list is ok. 14:19 doserj but server owners should have agree if their server is listed on a public list 14:19 PilzAdam there should be a GUI to store the server 14:20 doserj well, every server you entered once should be added to the local list (does it do that?) 14:21 PilzAdam yea, that would be nice 14:21 darkrose and an option to delete one from the list in the gui... that'd be good 14:22 doserj yes 14:22 darkrose and poeple can always share their server lists on the forum 14:22 PilzAdam maybe just a dropdown menu in the multiplayer tab that saves your servers and a "List" tab for online lists 14:23 doserj maybe make the local list your list of "favourite" servers, and have a way to populate it from a public remote server list? 14:24 doserj but definitely don't publish a list of servers without agreement of the server owners. that is just a big no-no. 14:25 darkrose the public list is a fail, imo, unless there's an official master server that the game server itself reports to on startup (unless a config setting says not to) to generate the list 14:26 darkrose otherwise it's just "servers that player X wants you to know about" which wouldn't be any different to sharing lists on the forums 14:27 darkrose ... and only using local files to populate the menu 14:28 PilzAdam just tested it and it crashed when pressing refresh 14:28 PilzAdam cant reproduce... 14:33 kotolegokot hmmmm, i have a segfault some time after the start of the game. gdb output: http://paste.ubuntu.com/1464612/ 14:34 hmmmm kotolegokot, is that from the version on the GIT? 14:34 kotolegokot hmmmm, yes 14:34 hmmmm mmm yes, in two calls to get_biome, the node_min.X and Y are passed instead of X and Z like it should 14:34 hmmmm so if you go too far in the Z direction it'll crash 14:35 hmmmm you'll see two lines like if (get_biome(blah, v2s16(node_min.X, node_min.Y)) { .... 14:35 hmmmm just need to change the Y to a Z 14:36 hmmmm mmmhrhmmmmmmgrmrg 14:36 hmmmm i need to push a fix for that 14:51 thexyz https://github.com/celeron55/minetest/pull/365 anybody? is there a point in making ipv6 disableable? 14:58 darkrose isn't IPv6 nonexistant on some windows versions? 15:01 thexyz probably in some 10-years-old versions 15:02 thexyz like winxp without any sp 15:03 darkrose if it's not disableable, you just know someone will bitch about it 15:04 thexyz well, at least, it should be made disableable in some other way 15:05 thexyz i don't see a point in https://github.com/celeron55/minetest/pull/365/files#L2R261, for example 15:25 doserj I guess he doesn't trust every isp to have their ipv6 configuration working properly 15:28 thexyz how's that related? 15:29 doserj maybe i misunderstood what you are criticising 15:30 celeron55 i don't see any point in making ipv6 disableable 15:31 celeron55 no other software does that... because there is no reason to 8) 15:32 celeron55 i'd say it needs to build and work in windows xp sp4 15:33 celeron55 maybe sp3 too if there is no real reason not to 15:34 celeron55 (or any version of xp for that matter; but anyway) 15:35 celeron55 also, yes, the server list needs to be able to store user-added servers from the UI 15:35 celeron55 independently of whatever else it does 15:37 celeron55 i guess somebody'd need to comment those pull requests on github as jeija and matttpt don't hang around in here 15:43 celeron55 hmmmm: the ipv6 addition could probably use testing on windows, can you do that sometime? https://github.com/celeron55/minetest/pull/365 15:49 hmmmm i stopped testing on windows, it's just not worth it 15:51 celeron55 crap, where is minetest.net once again 15:57 celeron55 well, dunno; depends on how much you think the 20000 monthly windows package downloads are worth 8) 16:09 hmmmm i used to have visual C++ 2008 all set up and working on my old computer through wine 16:09 hmmmm but i had to go through a lot of BS to get it working so this time around i just didn't bother 16:10 hmmmm instead now i have it set up on a virtual machine, but i had trouble the last time i tried to run any sort of 3d application in a vm 16:30 thexyz hmmmm: it works fine for me (virtualbox + minetest + nvidia blob) 16:33 hmmmm i'll take a look sometime 16:34 hmmmm are all versions of C++ okay with anonymous structs? 17:05 thexyz celeron55: i'm still open for hosting minetest.net 17:57 hmmmm i'm gonna be honest, over 65535 verticies for a single mapblock is crazy 17:57 hmmmm wouldn't this kill video performance? 17:58 hmmmm graphics 18:20 RealBadAngel im not sure how many vertices can have one single model in other games 18:21 RealBadAngel but i bet such buffers were thought for handling single objects not whole world look 18:21 VanessaE from what I read on irrlicht forums, this isn't uncommon. 18:22 VanessaE the solution presented was either to use a 32-bit count and dynamic buffers, or to break your work into smaller pieces each with their own 16-bit indexed buffer, e.g. head, torso, each arm, each leg in the case of a high-quality 3d player model 18:22 RealBadAngel well, simple cow model, 2k triangles, 1k vertices 18:23 RealBadAngel i havent heard of 65 cows killing the game performance 18:24 VanessaE so in the case of the trees, I could only say break them up by material type, then by height or something e.g. trunk, then leaves, then subdivide those into smaller parts. 18:24 RealBadAngel we will try to add dynamically just another one when needed 18:25 VanessaE yes, that's what c55 said he favors. 18:32 RealBadAngel lol, on the very same page as the cow i found A BUNNY 18:32 RealBadAngel 69k vertices :) 18:32 VanessaE heh 18:33 RealBadAngel i think its all about if over 64 can kill something 18:33 RealBadAngel 1 or 2 bunnies wont kill minetest for sure