Time Nick Message 00:02 sapier how do I enable fly mode? (didn't use it for some time) 00:02 VanessaE /grant singleplayer all 00:02 VanessaE (or your name) 00:03 VanessaE then hit 'k' to enable it and use shift + space to go down/up 00:03 VanessaE (or E + space if you use that) 00:04 sapier ok 00:04 sapier :-) starting takes ages with all that nodes in 00:05 VanessaE turn off item preload. 00:05 sapier can be done? 00:05 VanessaE Settings Menu, "Preload Item Visuals". Uncheck that. 00:05 sapier that's a candidate for a background thread 00:06 VanessaE yup 00:07 VanessaE but for now, having preloading turned off is good enough 00:08 VanessaE it also makes it possible for high-rez texture pack users to exit-to-menu and re-enter a world without having to restart the whole client. 00:08 sapier hmm I don't find anythin constructed where is everything? 00:08 VanessaE teleport to -138,24,100 00:08 VanessaE that's the spawn 00:09 sapier ok I assume wrong world :-) 00:09 VanessaE fly north-northwest a bit and you'll hit a heavily-wooded area 00:09 sapier north is? 00:09 VanessaE Z+ 00:10 VanessaE also southeast of the spawn is lots more wooded area. 00:10 VanessaE east = X+ 00:13 sapier how long does it take to crash? 00:16 VanessaE sapier: anywhere from a few minutes to several hours 00:16 sapier hmm not easy to test I'm flying above ungenerated are for some time now 00:18 VanessaE sapier: are you running my game with it? 00:18 sapier yes 00:18 VanessaE and are you flying through the heavily-wooded areas? 00:18 sapier yes already passed it 00:18 VanessaE singleplayer mode? 00:18 sapier no client server with item preload enabled 00:18 VanessaE hrm 00:18 sapier not quite sure if this makes a difference 00:19 sapier ok I've disabled preload and test again 00:19 VanessaE dunno 00:19 VanessaE preload is strictly a client-side thing though isn't it? 00:20 sapier i don't know 00:21 sapier ok preload or not doesn't seem to matter 00:21 sapier maybe you can check that additional commit 00:21 VanessaE you're running that commit? 00:21 sapier yes 00:21 sapier second one I posted 00:21 VanessaE ok, lemme try it 00:22 sapier it's experimental but if it works I'll finish it for merge 00:22 VanessaE is that first commit also needed? 00:22 sapier i think so yes 00:22 VanessaE ok 00:23 VanessaE building... 00:23 sapier I'm running on quad cor phenom ... so you should have even better performance 00:24 VanessaE ok, let's see if that works. 00:26 VanessaE hrm, I better re-clone. something isn't right here 00:26 sapier I'm still flying above woods 00:28 VanessaE there, re-cloned, got both of your commits, building system-wide. 00:29 VanessaE ok, let's see what this breask ;) 00:30 sapier I should be in bed for some time so plz make it quick ;-) 00:31 VanessaE heh ok 00:31 VanessaE gimme a minute or two 00:33 sapier I've just teleported to 5000,20,5000 directly into newly generated woods without crash 00:33 VanessaE ok, got the new build into place. 00:34 VanessaE it's still stalling out for me. 00:34 VanessaE though you don't have time for a full test so I can't say if it's going to completely hang up yet or not. 00:34 sapier hmm strange ... your server should outperform my pc ... do you have server and client on same machine? 00:34 VanessaE so far I just see the map loader taking several seconds at a time, for now. 00:34 VanessaE yes, they're on the same machine. 00:35 VanessaE vanessae.mine.bz:30000 if you wanna take a look around. 00:35 sapier thats even more strange ... what distribution are you at? 00:35 VanessaE Ubuntu 12.04/. 00:36 sapier ok I'm on wheezy 00:36 VanessaE it's lagging, but it *is* running 00:36 sapier another pthread version but that souldn't be such a huge difference 00:37 sapier ok I didn't expect performance to make huge jumps but only avoid emerge thread from beeing staled 00:37 VanessaE nope, it keeps stalling out 00:37 sapier hmm "waiting for content" 00:37 VanessaE I just waited about one full minute for the map to resume generating. 00:38 VanessaE initial media download is about 15MB so it'll take a while. 00:38 sapier there must be something else 00:38 sapier what kernel version is 12.04 running? 00:38 VanessaE vanessa@rainbird:~$ uname -a 00:38 VanessaE Linux rainbird 3.2.0-36-generic #57-Ubuntu SMP Tue Jan 8 21:44:52 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux 00:39 sapier debian too 00:39 sapier hexacore intel or amd? 00:39 VanessaE AMD Phenom II X6 1055T 00:40 sapier ok Im on phenom II x4 ... not a big difference 00:40 init wow VanessaE 00:40 init nice phenom :P 00:40 sapier is there any pattern like "occurs on ubuntu 12.04" only? 00:41 VanessaE sapier: beats me. 00:41 sapier no offence I just want to find out where the differences are 00:42 VanessaE none taken :-) 00:42 sapier wheezy and ubuntu 12.04 have different glibc thus different pthread and of course different locking 00:43 sapier it wouldn't be first time a glibc beeing buggy 00:43 VanessaE heh 00:44 sapier I can't join ... I'll try to reproduce your bug tomorrow at my machine I'll tell you if I succeed or not 00:44 sapier bye 00:45 VanessaE bye for now 00:45 VanessaE annnnd just as he leaves, my server hangs up. pff. 00:45 VanessaE there it goes. 00:53 VanessaE got logs... you know, I think that did improve things to some extent. 00:54 VanessaE not by much, but maybe a little 00:55 VanessaE oh wild 00:55 VanessaE I think the cave generator just tried to carve out a chunk of water in someone's waterfall. 00:55 init wtf? 00:56 VanessaE I just saw a glob of water ~5x5x5 nodes disappear and then the void started moving downwards 00:56 VanessaE (its a thick waterfall) 01:05 hmmmm you can set the scheduler type!? 01:05 hmmmm how on earth does that work 01:06 hmmmm are pthreads under linux apartment threads? 01:07 VanessaE dunno 06:16 celeron55_ < VanessaE> [01-16 17:28] could 06:16 celeron55_ they fix the bug where you have 06:16 celeron55_ to jiggle the mouse to open a 06:16 celeron55_ formspec after just closing one? 06:17 celeron55_ there is a patch for this made by kahrl, but i deemed it too large and messily designed back then 14:34 RealBadAngel hello minetest 15:31 dysoco Is the developer wiki down? http://minetest.net/wiki/doku.php 15:42 init dysoco: the only i know about it atm is that http://forum.minetest.net/viewtopic.php?id=4243 , but i'm not sure what happened with the old dev wiki 15:42 init souldn't it be online? 15:43 init and in dev.minetest.net i don't see info about the core 15:54 dysoco ah, http://dev.minetest.net/Main_Page 15:55 dysoco yes, it's fairly incomplete 16:59 celeron55 i couldn't get the old dev wiki to work 16:59 celeron55 when i transferred it to the current host of the main site 16:59 dysoco And the new wiki doesn't seem to have information about the Core. 16:59 celeron55 it just simply refuses to, showing nothing and no errors either 17:00 celeron55 this is since last week, so you're quite fast to complain 8) 17:01 celeron55 if anyone has experience of such behavior with dokuwiki, i'll have use for that experience 17:02 dysoco hehe, I've never been in the old wiki: I'm not complaining, maybe it was deprecated or something :P 17:02 celeron55 there's a bunch of relevant stuff in there that should be copied over to dev.minetest.net 17:02 celeron55 not much, but many pages 17:05 celeron55 but i'm simply helpless with this 17:06 celeron55 when dokuwiki shows an completely empty page, there is supposed to be some kind of a fatal error 17:06 celeron55 but even though i have logging enabled in php and apache, neither logs anything 17:06 dysoco apparently there is some kind of Documentation Sprint in 19. 17:06 celeron55 yes there is 17:06 celeron55 this would be very important for it too 17:07 dysoco Is the code well commented at least? I was thinking in reading through it. 17:07 doserj re liquid updates: I just realized that a map doesn't save the nodes to update over a shutdown. So when you stop and restart a server while liquids are updating, they get stuck. One more reason to put m_liquid_updates into MapBlock instead of Map, I guess. 17:08 doserj then Map can have a list of blocks instead of a list of nodes to update 17:09 celeron55 dysoco: many smart people have gotten to understand it; there'd be some useful info on minetest.net/wiki though... 17:10 celeron55 dysoco: the goal of saturday is exactly to document the C++ side of things to make it more accessible 17:10 dysoco I see, I'm looking to get more into Game Development and C++, but I really suck at design and I keep failing to do games by myself, so at least I'd like to get into an Opensource project. 17:10 doserj loading a block with liquid updates inserts it into the Map's list. and if the map finds a block in its list which is unloaded, it can safely remove it. 17:11 doserj still have to thinks about the case of mapclock borders to unloaded blocks, to make sure nothing gets forgotten there. 17:15 celeron55 dysoco: the general advice in game development seems to be to start very small and work up from there, doing increasingly sizeable projects 17:16 dysoco well, I have a Pong and a Space Invaders for Android, but still: the Art has been done with Paint :P 17:16 Calinou windows developer **alert** 17:16 Calinou ;-) 17:16 dysoco I mean, not paint 17:16 dysoco Paint = Not professional graphic editor. 17:17 dysoco not sure if it was Gimp or Inkscape 17:18 celeron55 as for how i did it, i made multiple very small games (just like those) many many many years ago, then did some small-ish non-game projects that i somewhat finished, started many medium-sized game projects that never got finished, contributed a bit to some non-game software, worked on some small-to-medium sized non-game software of my own and then started minetest... 17:19 dysoco yes, everything is possible with time and dedication :P 17:19 celeron55 art is quite a different thing compared to programming; you'll have to decide whether you want to be able to do it yourself (in which case, go draw stuff) or just specialize in programming (which makes it hard to make small projects) 17:20 celeron55 after starting minetest, i already have started many other projects of very different sizes :P 17:20 dysoco now that's my problem: I really suck at art, it always ****** my GPA in School :P 17:21 dysoco Anyways I need to leave now, I'll try to be here for the 19 to be what's going on. 17:22 celeron55 see you :P 17:22 * celeron55 goes back to fighting with dokuwiki 17:26 celeron55 omg 17:26 celeron55 it works when i removed the plugins directory 17:27 celeron55 apparently it was the columns plugin 17:28 * celeron55 throws it in the trash 17:28 Calinou "I really suck at art" => "Good Opensource Art: Last seen: Never" 17:28 Calinou :> 17:41 celeron55 thexyz: gimme edit privileges to dev.minetest.net main page 17:43 thexyz k 17:43 VanessaE "plz give me privs on ur server. ty." <--- what c55 said, in forumspeak ;-) 17:44 Calinou you forgot "i will not greif" 17:44 VanessaE right. 17:45 PilzAdam he didnt said that.... 17:45 VanessaE PilzAdam: fix your sarcasm detector. It's on the fritz again. 17:45 PilzAdam I mean the grief thing only) 17:47 VanessaE heh, right :) 17:54 celeron55 i wonder how we should differentiate the scripting API things from the core things 17:57 VanessaE no clue, except that it should be trivial to follow from some API bit over to the documentation for the part of the code that makes that bit work. 17:57 VanessaE (so really, kinda the opposite in a way) 17:57 celeron55 that is just links 17:57 VanessaE yep I know. 18:00 celeron55 anyhow, i added the relevant pages from the old wiki: http://dev.minetest.net/Main_Page 18:00 celeron55 well, about core, that is 18:01 celeron55 is this useful? i guess it could be added too, altough it's non-editable as-is: http://minetest.net/wiki/lib/exe/detail.php?id=code%3Ainner_workings_of_minetest&media=code:minetest-dfd-visio.png 18:02 PilzAdam isnt that 0.3? 18:02 celeron55 0.4 isn't much different to 0.3 18:03 celeron55 i'll copy it over too; somebody should probably redraw it (that is not made by me) 18:05 celeron55 (and i have no idea by who that is) 18:06 thexyz http://minetest.net/wiki/doku.php?do=recent&id=code%3Ainner_workings_of_minetest 18:07 celeron55 eh... okay; well, i do have his e-mail address then 18:07 thexyz it's also specified here https://github.com/janitor61/ 19:31 sapier vanessae do you have tohose strange to many objects in block messages too? 19:35 VanessaE sapier: do you mean the out-of-indexes problem that sometimes happens, or the "over 49 entities" issue? 19:39 sapier yes 19:40 thexyz wait, "yes" is not proper response to "or" question 19:41 sapier it is 19:41 sapier but it wont help very much i agree with that 19:42 sapier and yes was meant as yes for both 19:42 sapier vanessa how is plantlife doing spawning? 19:48 sapier vanessae ... I didn't have any crash when flying around in your world for about 20 minutes (with my patches applied) ... I did have stalls but only if flying throug woods where plantlive did grow or spawn flying through any other unloaded area didn't create any stall. 19:48 sapier is it supposed to crash? 19:51 VanessaE plantlife uses abms set to trigger on the presence of dirt with grass, but keeps a fairly sparse distribution of plants, usually only 2 or 3 per mapblock, and very seldom does the code that adds them actualy get to run 19:51 VanessaE ditto for trees. 19:51 RealBadAngel sapier, so you shall answer yes || yes 19:51 RealBadAngel ;) 19:52 VanessaE it doesn't so much crash as it hangs up, sometimes for hours on end. 19:52 VanessaE although I don't see it happening now since I added your patch 19:52 VanessaE I have to wait for others to tell me if it's still an issue 19:53 sapier yes || yes evaluates to yes which is a single answer too ;-) 19:53 sapier yes,yes using lua notification may be an option ;-) 19:54 sapier vanessae dirt and grass are often there even with low rates they do make significant load 19:55 VanessaE yes, but when this happens, I can usually still chat, send commands, etc. 19:55 sapier especially on chunk loading as all abms start with less distribution 19:55 VanessaE if it were an ABM hanging the server, those would not work either. 19:55 RealBadAngel sapier 4 can be an answer but you dont know if the question was 3+1 or 2+2 ;) 19:55 VanessaE bbiab 19:57 sapier I don't think so abm's not quite stall it completely but chung loading and generating is far to vulnerable to disturbances introduced by lua 19:57 sapier rba don't know what you're trying to tell me ;-) 19:58 RealBadAngel cave gen over ground introduces HEAVY lags to mapgen 19:59 RealBadAngel caves are generated all over, and no matter how high 19:59 sapier node setting in on_generated is even worse ... and abm's on default nodes should be forbidden ;-) 19:59 sapier is this done by lua or core? 19:59 RealBadAngel simply disabling cave gen with height limit speeds mapgen greatly 20:00 sapier vanessae if those patches proove to improve situation about some time I'll finish them for merge 20:00 RealBadAngel i posted once a dirty fix for that 20:00 sapier whats root cause for cavegen to behave that bad? 20:00 RealBadAngel bad code 20:01 sapier that answer is as good as yes ;-) 20:01 RealBadAngel caves are generated on every height 20:01 sapier yes but that shouldn't be a problem 20:01 RealBadAngel it is, wanna SEE it? 20:01 sapier except of strange maps of course 20:01 sapier but not for performance 20:01 RealBadAngel hold on 20:02 sapier I've already seen those pictures with cave/mese clouds in sky ;) 20:04 RealBadAngel ah, ok 20:04 sapier but still this schouldn't have such big performance impactr 20:04 sapier -r 20:04 RealBadAngel wanna bet? 20:04 RealBadAngel they do 20:05 RealBadAngel caves are simply everywhere 20:05 sapier if you disable 1/3 of an algorithm its faster per definition ;-) but cavengen taking 1/3 of load is already a problem 20:05 RealBadAngel even on y=+10000 mapgen lags 20:05 RealBadAngel because of caves 20:05 sapier don't know if 1/3 1/2 or 2/§ is correct 20:06 sapier yes but disabling isn't a solution but a workaround 20:06 RealBadAngel more, it causes lighting bugs (dark trees) 20:06 RealBadAngel floating island (thx to blobs in caves) 20:06 RealBadAngel trees in the sky 20:07 sapier still disabling ain't a solution ;-) solution would be fix it's performace footprint and have it run within solid ground only 20:08 RealBadAngel solution is quite simple 20:08 sapier I do listen? 20:08 RealBadAngel hmmm figured it out 20:08 RealBadAngel avg height of the block calculated and compared to max_y of generated block 20:09 RealBadAngel if max_y is out of bonds then skip cave code 20:09 sapier average height? I hope calculated in core but not lua? 20:10 RealBadAngel avg is taken from perlin noise for given x and z coords 20:10 RealBadAngel we are talkin c++ all the time 20:10 sapier are we talking about blocks or nodes? 20:10 RealBadAngel and blocks 20:10 sapier ok so average can be calclulated relatively simple 20:12 RealBadAngel i got it somwhere, lemme find it 20:12 sapier sounds like a good solution 20:13 RealBadAngel https://github.com/celeron55/minetest/blob/master/src/mapgen.cpp#L1346 20:13 RealBadAngel theres start of cave code 20:13 sapier but would you get entrances to caves? 20:13 RealBadAngel yes 20:13 sapier only if a block ends "on surface"? 20:14 RealBadAngel http://pastebin.com/6vbfai4k 20:14 RealBadAngel replace the start with it 20:14 RealBadAngel thats a major fix for the cave gen 20:14 RealBadAngel skip air cave generation 20:15 RealBadAngel caves are still generatin in air below avg height 20:15 RealBadAngel in mountains, cliffs etc 20:16 RealBadAngel but you will notice great speed up of terrain generation for sure 20:17 RealBadAngel of course only on the ground 20:17 RealBadAngel not deep down 20:18 sapier it's an improvement but I'm not quite sure how this behaves with different block sizes 20:18 RealBadAngel blocks dont have different sizes afaik 20:19 RealBadAngel its fixed 20:19 sapier if on_generated uses same sizes as blocks they do? it's fixed? 20:19 RealBadAngel 16x16x16 20:19 celeron55 ha! 20:19 sapier what does on_generated run at? 20:19 sapier hello celeron 20:19 RealBadAngel am i wrong c55? 20:20 celeron55 RealBadAngel: use maximum height, not average height 20:20 RealBadAngel ok 20:20 RealBadAngel but i will need to add more check points 20:21 RealBadAngel maybe in cross shape to detect cliffs 20:21 celeron55 well, you'd need that for good average too 20:21 celeron55 (equally much) 20:22 RealBadAngel 5 points out of 16x16 would be fair enough 20:22 RealBadAngel center and 4 extra in middle of each direction 20:23 celeron55 and the generator currently generates areas of 5 blocks to every direction at a time, not a single one 20:23 celeron55 80x80x80 areas, that is 20:24 sapier rba I'm not sure but randomly selecting points might give better results 20:24 celeron55 or, volumes, to be exact 20:24 sapier are those cubes directly transmitted to on_generated lua call? 20:24 celeron55 yes 20:24 RealBadAngel im not sure if random will be any good there 20:24 RealBadAngel its an average 20:25 sapier strange I'm quite sure I got different values for min_p max_p 20:25 sapier might be a bug somewhere else 20:25 RealBadAngel middle points shall work the best to calculate average not random ones IMHO 20:25 sapier having a fixed pattern will make you prone to some strange patterns that may be introduced by trees for example 20:26 celeron55 RealBadAngel: at least i don't care if you just pick the corners, but you could divide the area into a grid and sample that way too; whatever 20:26 RealBadAngel i said: middle 20:26 RealBadAngel not borders 20:26 sapier doesn't make a difference if middle or border 20:27 RealBadAngel if the terrain is continuous it makes 20:27 sapier no ... it's only a quantitative difference 20:27 celeron55 oh god 8D just do something; as long as there are multiple points and you choose the maximum, it's fine 20:27 sapier quality stays same 20:28 RealBadAngel 5 points shall do the job 20:28 RealBadAngel no matter borders or middle 20:28 sapier still I don't want to calculate how many random points you need to select to get a accuracy of XX% 20:29 RealBadAngel we wont hit perfection with that for sure 20:29 sapier of course it most likely will be good enough 20:29 RealBadAngel but definitely we will get closer to it ;) 20:30 RealBadAngel btw 20:30 sapier celeron is SCHED_FIFO only because it's default scheduling policy or is there any reason for it? 20:30 RealBadAngel i noticed some SPIKES in terrain gen 20:31 celeron55 sapier: what is that 20:31 RealBadAngel like single stone goin up to the sky 20:31 sapier pthread scheduling policy? it's related to how threads are scheduled e.g. on release of mutex 20:32 celeron55 i have no idea of anything related to that 20:33 RealBadAngel celeron55, could it be some perlin bug or some rounding issue? 20:33 sapier vanessae is trying some experimental patches I created yesterday to avoid emerge thread from stalling completely ... major changes have been lift emerge thread priority and switch fo sched_RR (round robin) 20:33 celeron55 RealBadAngel: it's not a bug; it's just what the algorithm can sometimes end up making 20:33 sapier SCHED_RR often improves responsiveness of different threads with same priority on heavy load 20:33 celeron55 it's essentially equal to floating islands in a 3D noise mapgen (like in 0.3) 20:34 RealBadAngel so maybe add some spike detection there? 20:34 sapier why? 20:34 RealBadAngel +20 nodes stone spike looks at least weird 20:35 sapier there are stone spikes in some real world areas too? 20:35 hmmmm it's better to have buggy, interesting terrain than correct, boring terrain 20:35 RealBadAngel buggy and smooth are two differnt things 20:36 sapier world isn't smoth everywhere too ;-) 20:36 hmmmm if you can find the reason why it makes those spikes, then sure, fix it..... but it's not really a priority at all 20:36 RealBadAngel have you seen 1x1 meter and 20 meters high pillar of stone IRL? ;) 20:37 hmmmm actually i don't think minecraft nodes are a cubic meter. they look quite a bit smaller than a meter in gameplay. 20:37 sapier http://de.wikipedia.org/wiki/Obelisk_von_Luxor 20:37 sapier :-) 20:37 RealBadAngel lol 20:38 RealBadAngel do we have some ancient egyptians messing with the code? 20:38 sapier add spike detection adding hiroglyphs ;-) 20:38 RealBadAngel rotfl 20:39 RealBadAngel that could be even a nice idea 20:40 RealBadAngel add some monumental ancient buildings 20:40 sapier let things like that do lua mods 20:40 RealBadAngel sure, not the core 20:42 sapier still modifications like that done in on_generated are not quite performant 20:43 hmmmm i think i should change that 20:43 sapier what do you mean with "that"? 20:44 RealBadAngel he means propably after 2 weeks of not codin "i will code something" ;) 20:44 hmmmm i have an idea. the engine passes a MapBlock to on_generate, the lua code directly modifies it, returns 1 if anything in the block was modified (and requires blitting back) 20:45 hmmmm as i understand right now it calls set_node for every single node 20:45 hmmmm which is horrendous 20:45 sapier true 20:45 RealBadAngel ah, you remind me of somethin 20:46 RealBadAngel i got the blocks in the "cache" i believe 20:46 RealBadAngel i put there my tree 20:46 thexyz hmmmm: what did you want to talk about? 20:46 hmmmm leveldb 20:46 RealBadAngel and checkin before puttin another tree node whats there 20:47 RealBadAngel problem is my changes do not affect the checks 20:47 RealBadAngel so if i put tree trunk there, check whats there, it says "air" 20:48 RealBadAngel so treegen replaces trunks with leaves 20:48 thexyz hmmmm: so? anything wrong with it? 20:48 hmmmm well i want to find out 20:48 hmmmm does your modifications break compatibility with older map formats? 20:48 hmmmm s/does/do/ 20:49 thexyz dunno about sectors format, but it works fine with sqlite3 20:49 thexyz https://github.com/minetest/minetest/commits/db_backends_2 20:49 hmmmm oh nevermind 20:50 hmmmm leveldb is SQL? 20:50 hmmmm for some reason i thought it was a key/value no-sql type database 20:50 thexyz it is 20:50 thexyz key-value 20:50 thexyz not sql 20:51 sapier can someone plz break down all those unnecessary big locks within minetest prior adding other heavy weight additions? :) 20:51 thexyz you are free to choose which backend to use per-world 20:51 hmmmm what the heck, github is not reponding 20:51 hmmmm they have some cruddy JS 20:51 thexyz it also can "migrate" from one backend to another 20:52 hmmmm great 20:52 hmmmm so everything is basically in place 20:52 hmmmm would there be any additions needed for this to be released to the general public? 20:53 thexyz well, things usually start to break one released 20:53 thexyz ATM it works fine at all OldCoder's servers 20:54 thexyz *once 20:54 hmmmm right, but things will never go anywhere if it's not released 20:54 hmmmm i was talking to celeron about it and we think that the leveldb stuff should go into master asap 20:55 sapier minetest is betatesting anyway ;) 20:55 thexyz well, the one (huge?) problem with leveldb is that it's tricky to compile under windows 20:55 hmmmm ..hah. that's a good point i never really considered.. 20:55 RealBadAngel leveldb on 19 worlds is workin just fine 20:55 hmmmm tricky? 20:55 hmmmm explain more 20:56 thexyz well, it seems it only compiles with boost 20:56 RealBadAngel and solved world corruption issues 20:56 hmmmm what happens with stl? 20:56 thexyz although there is some patch which makes it possible to compile without boost, dunno if it works 20:57 thexyz hmmmm: what should happend with it? 20:57 thexyz I've never tried to compile it though 20:57 sapier boost? 20:57 thexyz there's -DENABLE_LEVELDB cmake setting 20:57 hmmmm do you have access to a windows environment to compile with? 20:57 thexyz yes, I do 20:57 sapier does this mean minetest will require boost soon? 20:57 hmmmm would you at least try to see what the problem would be ?:) 20:58 hmmmm sapier, of course not 20:58 sapier puuuuh 20:58 sapier that'd be a reason for forking ;-) 20:58 hmmmm we're either going to fix this directly or check out that "patch" that fixes it 20:58 hmmmm but if it comes down to that 20:58 hmmmm we might need to just for the windows builds 20:59 hmmmm there aren't too many windows minetest users though, and it'll only be necessary for that small subset of people 21:00 sapier windows users are like binary package linux users ... rare but important 21:00 thexyz https://groups.google.com/forum/?fromgroups=#!topic/leveldb/Y7eoVJdS4pY 21:00 sapier at least if goal is to spread minetest 21:01 celeron55 w-what? :D not many windows minetest users? 21:01 celeron55 the situation is: most of minetest users use windows, most of minetest developers use linux 21:03 sapier sounds reasonable 21:03 thexyz whatever, windows minetest users don't usually build packages 21:03 RealBadAngel but the community needs simple users too. that means windows. 21:03 RealBadAngel click, download and play 21:04 sapier so windows build shouldn't be too complicated ... it's already complicated 21:04 RealBadAngel windows compile is pain in the ass 21:05 RealBadAngel and it was one of the main reasons i finally deleted windows 21:05 thexyz sapier: that doesn't matter, *we* build windows packages, not users 21:05 celeron55 in my viewpoint, the windows build should be simple enough to: allow there to be multiple people who are able to and care to build dev builds, and allow there to be multiple people who are willing to test new things on windows 21:05 celeron55 it is *not* enough if only a build is made for releases 21:06 celeron55 it will end up in chaos 21:06 sapier as you will find windows only bugs after release 21:06 celeron55 yes, and the windows build would rot in all imaginable ways to be unusable in the end 21:06 RealBadAngel celeron55, typical luser when seen "build" "compile" will turn away 21:06 RealBadAngel because he simply dont get it 21:06 celeron55 RealBadAngel: ... 21:07 celeron55 RealBadAngel: are you sure you cannot understand what i said? 21:08 celeron55 hmmmm, thexyz: i have recently stumbled (in work) to a project that uses boost, that includes boost as a git submodule 21:08 celeron55 it might be an option 21:09 RealBadAngel i can, but i wanted to higlight my point of view. i get lotsa money each month for installing 7zip or other basic stuff for folks 21:09 sapier to get a windows build env took about half a day and I'm used to compile on windows AND linux so even current way of doing is complicated 21:09 celeron55 i haven't tried building that project on windows yet though 21:09 thexyz celeron55: the boost is only needed for win builds 21:09 RealBadAngel there should be just one link: download, play 21:09 celeron55 thexyz: submodules are not fetched unless you specifically ask git to do it 21:10 thexyz i know, i just think it's not needed 21:11 thexyz when we don't include, say, irrlicht as submodule, what's the point of including boost? 21:11 RealBadAngel take a look on mentioned github, folks even have problems to rename the mod folder 21:11 thexyz by the way, RealBadAngel, what's the problem you're talking about? 21:11 thexyz i don't get it 21:12 sapier it's about lifting level to enter minetest development 21:12 celeron55 RealBadAngel doesn't seem to understand at all what we are talking about 21:12 sapier it's already very high 21:12 RealBadAngel heh, ive tried to hit a different shelf 21:12 celeron55 thexyz: is there a ready-to-use boost distribution for windows like there is of irrlicht? 21:13 thexyz sapier: it's only high for windows devs 21:13 celeron55 also, zlib could be included in the source like jthread is included 21:13 thexyz how many windows devs have we got? 21:13 thexyz celeron55: dunno 21:13 sapier step back, have a look at minetest code and you'll see it is high ;-) 21:13 RealBadAngel nekogloop ;) 21:13 celeron55 i stand with what i said 8 minutes ago 21:13 thexyz i mean, core devs 21:13 RealBadAngel and Jordach propably 21:14 celeron55 i haven't yet seen a good argument against it 21:14 RealBadAngel for me as i said MT was good enough reason to delete windows partition ;) 21:17 RealBadAngel but, if you all havent got what i was tryin to say 21:17 RealBadAngel if newcomer finds out about Minetest 21:18 RealBadAngel fires up google, types minestest download 21:18 RealBadAngel wants to get it and play 21:18 RealBadAngel he gots nothin 21:18 thexyz wait 21:19 thexyz he gets here http://minetest.net/download.php 21:19 sapier I do get what you mean, those users are important too 21:19 sapier but those willing to support development are even more important 21:19 celeron55 RealBadAngel: nobody has questioned such users *at all* 21:19 celeron55 RealBadAngel: stop making up problems out of nothing 21:20 RealBadAngel im not tryin to make problems 21:20 RealBadAngel i know typical users just 21:21 thexyz i still don't get it 21:21 RealBadAngel you, as a coder can just not get it 21:21 celeron55 so, what happens then when thexyz isn't around anymore and nobody wants to build new minetest versions for windows? 21:21 thexyz why can't that user download http://minetest.net/download.php 21:21 RealBadAngel because some things are obvious for ya 21:22 celeron55 and i am not around either 21:22 celeron55 the linux people will fire up their virtual machines and look up the instructions and say "bleh", and then the community just sunk to 30% size 21:22 thexyz oh 21:23 RealBadAngel because: download, rename, where the fuck are mods, rename them, error, wtf, im off 21:25 celeron55 knowing what the communities of easy-to-use free-to-play online multiplayer games are like, i'm not exactly eager to get the same people to use minetest 8) 21:26 celeron55 we don't have need for that kind of people, because we don't monetize them 21:26 celeron55 monetization is the only reason for that 21:26 RealBadAngel take this, i tried to "sell" minetest to 9 yrs old kid from my family 21:26 RealBadAngel which is playin Minecraft all the time 21:27 RealBadAngel when i started to explain him what he has to do to play it 21:27 RealBadAngel i just gave up at some point 21:28 RealBadAngel read the forums, look for most "not workin" posts 21:28 celeron55 nobody is prohibiting anyone from making an installer and a website with a single button to launch it 21:29 RealBadAngel YES 21:29 sapier still i understand rba's point especially a mod package format supported by core would be interesting 21:29 sapier but I think someone is working on this is that true? 21:29 RealBadAngel martraceur has shitload of rigjht right and anc full TIR of right 21:29 celeron55 i don't think anyone is interested in doing this 21:30 RealBadAngel he is 21:30 celeron55 also i think RBA is pretty drunk 21:30 RealBadAngel no, was typin too fast ;) 21:30 * marktraceur wonders what's up here 21:30 sapier cheers 21:30 marktraceur RealBadAngel: I don't know what you're saying above 21:31 RealBadAngel make that mod manager 21:31 celeron55 nobody else knows either, and all the wise people have already shut up :P 21:31 marktraceur Ah. 21:31 marktraceur celeron55: "Package management" is on my list somewhere between "bug triaging" and "code sprint", probably 21:31 RealBadAngel and somebody else easy to find builds 21:32 sapier bugfixing is more important 21:32 RealBadAngel and third make them "download" and "play" 21:32 marktraceur "Continuous integration for builds" probably comes in after "code sprint" but it's really not even on my radar 21:32 hmmmm speaking of sprinting, i do intend to add to the documentation sprint 21:32 marktraceur celeron55: First step would be to get a jenkins instance up and running 21:32 sapier celeron is there any chance to get something like a "functions have to be less than 100 lines of code" rule to minetest? 21:32 hmmmm but right now i am sprinting to type up this feature list 21:32 marktraceur (or you know, whatever CI thing you want) 21:33 marktraceur sapier: Most wouldn't make that cut 21:33 * celeron55 hides (don't ask me to do anything) 21:33 RealBadAngel hehe 21:33 sapier 500+ loc functions are monsters 21:34 hmmmm yeah :/ 21:34 hmmmm "there is no reason to put things in separate functions if they are only called once!" 21:34 RealBadAngel celeron55, btw i fixed indices problem, testin that out since a few days 21:35 sapier there is one big reason for this ... you can imagine what a speaking function does but you can't remember what 300lines of code do 21:35 hmmmm (translated, this means "i don't feel like encapsulating the relevant variables into a class, and C++ doesn't have dynamic scoping" 21:36 hmmmm i intended to turn the_game into a class a long time ago and breaking it up, but i gave up after a couple days. 21:36 sapier I still don't get a real reason from what you're telling? 21:37 sapier it's not only the_game 21:37 hmmmm i know, it's lots of other things 21:37 hmmmm don't forget a lot of that is the pseudo-GNU code style and comments and whitespace 21:37 marktraceur celeron55: Jenkins isn't terribly impossible to set up, but if you'd prefer not to deal with it right now, thexyz and I can collaborate later maybe. 21:37 thexyz marktraceur: regarding CI, some github projects use https://travis-ci.org/ 21:37 marktraceur This I know 21:38 sapier e.g. client I had a look at getNextBlocks yesterday I've been scrolling up and down for some time to really understand what's going on there 21:38 hmmmm i would really prefer if the code style in minetest were something like the Linux kernel style 21:38 marktraceur thexyz: I'd almost certainly prefer having our own infrastructure. 21:38 RealBadAngel and i wish all tabs and spaces would be automagically deleted when posting ;) 21:39 thexyz so let's start by switching from github to self-hosted gitorious?) 21:39 marktraceur thexyz: Don't think that's not on my list :P 21:39 * marktraceur is mostly kidding 21:39 sapier let's start by categorizing github issues ;-P 21:39 marktraceur sapier: That will happen soon :) 21:40 sapier VanessaE will you tell me if those experimental changes do have a positive effect? 21:40 marktraceur Speaking of, really helpful thing for that purpose and, ostensibly, for the doc sprint (or parts of it): Could I have access to the GitHub bug tracker so we can tag/close bugs? 21:40 marktraceur celeron55, thexyz (maybe), and others ^ 21:41 marktraceur I won't presume to ask for commit access, but bug tracker access would be useful 21:42 celeron55 there is only a single "collaborators" list 21:42 celeron55 either people can do anything, or nothing 21:42 marktraceur Hm. 21:42 celeron55 well, not anything; but whateve the maximum extent of grantable things is 21:42 celeron55 +r 21:42 marktraceur celeron55: Well, FWIW, I don't intend to commit anything. Self-review is bad. 21:42 celeron55 the github issue tracker is quite minimalistic 21:43 marktraceur And reviewing others' stuff, is not something I'm qualified for, so I won't. 21:43 marktraceur (at least not for now) 21:43 celeron55 i guess i can add you... and while at it, i'll add hmmmm too 21:43 hmmmm hmmmm 21:43 celeron55 just, like, don't mess things up 21:44 hmmmm okay i'll be sure to screw it up as fast as possible 21:45 RealBadAngel celeron55, can you help me with one issue with treegen? 21:47 celeron55 (and by not messing things up, i don't mean not doing anything; doing nothing is messing things up too, because things get messed up if nobody does anything!) 21:47 hmmmm by the way i almost have that typed up 21:47 hmmmm i'm still working on it, not just sitting around doing nothing 21:48 marktraceur Solemn oath 21:49 marktraceur I, state your name, promise not to fuck it up. 21:50 RealBadAngel celeron55, ? 21:50 celeron55 also, if people still want to strictly follow what i say, i must say that having a completelyu horrible windows build is better than stopping development 21:50 celeron55 -u 21:51 celeron55 but be aware that it is a change to what we currently have, and may have unpredictable consequences! 21:51 RealBadAngel Having Minetest is way better than Minecraft for sure ;) 21:52 celeron55 this is a game after all, not the regular nerdy project like vim or nginx 21:52 RealBadAngel but please, if you can try to explain me code behaviour :) 21:53 celeron55 RealBadAngel: not now; i'm going to sleep 21:53 RealBadAngel its not a big stuff 21:54 RealBadAngel but annoying 21:54 hmmmm if i've learned one thing, it's that i shouldn't stay up later than usual for things on IRC 21:54 marktraceur hmmmm: But people are WRONG on the INTERNET! 21:54 hmmmm none of that 4AM crap, seriously 21:54 RealBadAngel on the other hand weekend starts tommorow 21:55 hmmmm well, it's midnight for him, but he probably has to get up early. 21:55 RealBadAngel celeron55, i will hunt you down for this later on then ;) 21:55 RealBadAngel i think its a bug 21:56 hmmmm what is it, maybe i can help 21:56 RealBadAngel ok, listen then 21:56 RealBadAngel when you want to emerge the map 21:57 RealBadAngel you define the range for changhes 21:57 thexyz hm, https://travis-ci.org/minetest/minetest/builds/4218594 21:57 VanessaE sapier: well the server is still lagging out sometimes, but it is sort-of responsive now. It's hard to say for sure, but I think your changes helped a bit. 21:57 VanessaE at the very least, they don't seem to have caused any regression 21:58 RealBadAngel then your code makes the changes to cached area, which is supposed to blitz back later on 21:58 RealBadAngel and be synced with the map 21:58 RealBadAngel do you follow? 21:58 sapier vanessae that's best I was hoping for as changes aren't that big 21:59 hmmmm yea 21:59 RealBadAngel now, the problem 21:59 RealBadAngel if your code makes changes to the cached area 21:59 RealBadAngel and checks for contens 22:00 RealBadAngel it doesnt see your changes 22:00 RealBadAngel but the original look of it 22:00 hmmmm can i see the code where you do this? 22:00 RealBadAngel treegen 22:00 hmmmm hold on 22:01 RealBadAngel leaves code is kinda messy 22:01 RealBadAngel and replaces already placed trunks 22:02 RealBadAngel checks on the coords doesnt help 22:02 hmmmm spawn_ltree? 22:02 RealBadAngel if the trunk was placed there 22:02 RealBadAngel and then leaves code checks whats on this coords 22:03 RealBadAngel it doesnt see the trunk placement 22:03 RealBadAngel it says "air" or whatever was there before 22:04 hmmmm hmm 22:04 RealBadAngel like the check was not reading the cached area 22:04 RealBadAngel but the original one still 22:05 hmmmm this doesn't happen with normal trees does it 22:05 RealBadAngel those are too small propably to see the problem 22:06 hmmmm you read it directly from vmanip.m_data[], no? 22:06 RealBadAngel yes 22:06 hmmmm that's just plain data 22:06 hmmmm there's no way anything else is changing it back to anything else 22:06 RealBadAngel i get it in vmanip and operatin only on it 22:07 RealBadAngel and then push it all back to map 22:08 RealBadAngel but the check seems to be readin the map not the vmanip 22:09 hmmmm this isn't the problem, but something i noticed 22:09 RealBadAngel no matter what you put on the cooords, check will return the state before gettin vmanip 22:09 hmmmm in make_tree_leaves_placement and others you have the index vi 22:10 hmmmm but you only use it to compare the content ids to CONTENT_AIR and CONTENT_IGNORE 22:10 hmmmm and you go back to using vmanip.m_area.index(p1) for the rest of them 22:10 RealBadAngel i got another code local 22:11 RealBadAngel i tried to fix it and testin coords for trunks 22:11 hmmmm oh wait a minute 22:11 hmmmm maybe i am misunderstanding 22:11 RealBadAngel even if i KNOW theres a trunk check pass it by with air 22:11 hmmmm is it replacing the tree trunks of OTHER trees that were already spawned 22:11 hmmmm ? 22:12 RealBadAngel no, its the same code 22:12 hmmmm if that's so then it has to be a problem with your code 22:12 RealBadAngel i pick vmanip, first put trunks 22:12 RealBadAngel then leaves 22:13 RealBadAngel leaves dont see trunks placed in it 22:13 hmmmm definitely there's a problem on your end 22:13 hmmmm what it is, exactly, i am looking for 22:14 RealBadAngel it may happed (thx to rounding, and complex calculations) that leaves wants to be spawned on the very same coords that the trunk was already spawned 22:15 hmmmm yeah but that doesn't matter because it checks before it sets. 22:15 RealBadAngel actual code checks if theres and air there to proceed, yes? 22:15 RealBadAngel or ignore to skip 22:15 hmmmm no, to skip if it's not air and not ignore 22:16 hmmmm it's logically equivalent to 'proceed if air' and 'skip if not air' 22:16 RealBadAngel skip if not air was intended 22:17 hmmmm i don't know, i can't see any problem 22:17 hmmmm i'll need to take a better look sometime 22:17 RealBadAngel i replaces already placed trunk nodes 22:17 RealBadAngel *it 22:17 RealBadAngel as if they were air 22:18 RealBadAngel and checks returrns "air" result 22:18 hmmmm from what i'm seeing, if you're having a problem with anything at all, it would be that the leaves from a new tree overwrite the branches of an old tree 22:18 RealBadAngel i dumped them on the screen 22:18 hmmmm there's just no other way anything else can happen 22:19 RealBadAngel i will try tommorow to reproduce the effect with simpler code 22:19 hmmmm okay 22:19 hmmmm why not try running it in a debugger? 22:20 hmmmm step through a little 22:20 RealBadAngel i think i will have to 22:21 RealBadAngel 20 years ago i had one VGA to run Diablo and second was hercules with Soft-Ice with debugger 22:22 RealBadAngel i killed diablo in my students hostel in like 2 weeks 22:22 hmmmm hey, just think, since minetest isn't fullscreen it's that much easier 22:23 RealBadAngel all were having maxed out weapons and stuff ;) 22:23 hmmmm to save yourself frustration, check out 'GDB TUI' :) 22:23 RealBadAngel cmon, 2x22'' here 22:23 RealBadAngel i do have a workplace ;) 22:24 hmmmm oh yeah? well i have THREE 22:24 RealBadAngel im not such rich yet ;O 22:24 RealBadAngel but two are just comfortable 22:24 hmmmm me neither.. dell 19" was $19, emachines 23" was $120, emachines 19" was $90 22:25 hmmmm that's a total of 229 USD on monitors 22:26 hmmmm some people blow away more than that for a single monitor..... i probably will never understand. 22:26 RealBadAngel ive seen a vid of some folk playin new diablo on 3 40+ inches 22:26 RealBadAngel amazing 22:28 RealBadAngel http://www.youtube.com/watch?v=8BveNnv0mrU 22:52 thexyz oh, great, https://travis-ci.org/minetest/minetest/builds/4218594 22:53 VanessaE er... 22:54 thexyz VanessaE: ? 22:54 VanessaE nothing, nevermind. 22:55 thexyz celeron55: shall we integrate with travis? 23:00 thexyz here's working .travis.yml https://raw.github.com/minetest/minetest/travis/.travis.yml