Time Nick Message 00:54 PilzAdam hmmmm, isnt it kinda inconsistent to have different ways of not setting a node in a schematic in the raw or file format? 00:54 hmmmm if you can come up with a way to fit more than 256 values in 8 bits, please do so 01:16 VanessaE hmmmm: the default is fff = 4096, how's that related to an 8-bit number? 01:17 VanessaE also, I've been running this on my machine and so far, it works perfectly other than unmodified clients being unable to connect. 01:17 hmmmm vanessae, we were talking about something different 01:17 VanessaE oh ok 01:18 VanessaE oh hell 01:18 VanessaE I missed a couple of words in there. 01:18 VanessaE anyway, that aside, it works perfectly fine here. 01:31 VanessaE hmmmm: khonkhortisan is on my server now with that same change. he says it's working fine for him also. 01:32 hmmmm yes, i am not doubting that it's going to instantly break 01:32 hmmmm that it won't* 01:33 VanessaE what will likey break? 01:33 hmmmm that's not the point though. the point is that simply changing it is doing it the wrong way 01:33 VanessaE likely* 01:33 hmmmm i am not doubting that it works okay* 01:33 VanessaE ah, that reads better. 01:33 hmmmm sorta distracted 01:33 VanessaE s'ok 01:34 VanessaE I can understand if that one-liner was the "wrong" way to fix the problem, but it *does* fix it 01:34 VanessaE can't a concession be made to do it the "wrong" way now and put in the "right" way some time later on? 01:34 hmmmm the thing that's broken here is having so many nodes and they're merely 100 different colors of the same exact thing 01:34 hmmmm no, because nobody ever does.. 01:35 VanessaE *sigh* 01:35 VanessaE as for 100 different colors, actually no. 01:35 hmmmm ah yes, 89 if i recall 01:35 VanessaE sure there are tons of colored blocks 01:36 VanessaE but that's not the issue now 01:36 VanessaE in fact what pushed it "over the line" if you will, was a mod that adds two new pipeline types to my pipeworks mod. 01:36 VanessaE s/pipeline/tube circuit/ ... they're color-coded sure, but originally they were gonna be called things like "cobble tube" and so on. 01:37 hmmmm it just seems like there are 100 minor variations of every node featured in every mod 01:37 hmmmm like, "oh, but this one's a left-right-top corner connector!" 01:37 VanessaE that's because there's simply no better way to do it 01:37 hmmmm "oh, but this one connects top left and bottom middle" 01:37 VanessaE not if you're talking about pre-existing code anyway 01:38 khonkhortisan "oh, but this one can connect left-right, forward-back, or both at the same time!" 01:38 hmmmm :/ 01:38 VanessaE (6d facedir can help to some degree, but in my install it would only trim a couple hundred nodes off the total) 01:38 hmmmm now how is it that minecraft can do so much with less than 256 nodes 01:38 khonkhortisan I made that one, the crossover soon-to-be in mesecons 01:38 VanessaE I think minecraft does more than that. 01:38 khonkhortisan redstone makes there be more nodes 01:38 VanessaE someone said 4096 was their limit also 01:38 khonkhortisan or dust? 01:39 khonkhortisan the one that gets added to the surface 01:39 hmmmm i mean the amount of nodes they actually have registered 01:39 hmmmm and people seem more than happy with that 01:39 VanessaE khonkhortisan: that's just "redstone" I think. 01:39 khonkhortisan the one you make a wire out of, is that added to what you put it on, or is it in the air? 01:39 hmmmm I think it would be worthwhile to have some sort of code that makes all different permutations of nodeboxes 01:39 VanessaE khonkhortisan: I think it's a dust when it's still in-hand. 01:40 kahrl yes, MC's limit is 4096, it was 256 before Anvil 01:40 VanessaE hmmmm: define "all different permutations" 01:40 khonkhortisan stairs + wires 01:40 VanessaE nodeboxes are floating point 01:40 hmmmm or like fences 01:40 VanessaE there's no such thing as "all possible permutations" 01:40 hmmmm i mean all different permutations of right right left, right left, top corner, right corner, lower right corner, blah blah 01:40 VanessaE oh that 01:40 kahrl the difference with MC is that mods can redefine the rendering code easily, so they can do fancier stuff with their versions of param1/param2 01:40 VanessaE no thank you. 01:40 hmmmm i mean, i can't help but look at that and think to myself, how absurd 01:40 VanessaE pipeworks already has that area cornered. 01:41 VanessaE as does mesecons. 01:42 VanessaE (and technic, and half a dozen other mods that have to use similar tricks) 01:43 VanessaE I say raise the limit. if someone hits it again at 64k, THEN we can get into discussions about what is or isn't the best way to code a mod :) 01:43 kahrl eh 01:43 kahrl and you're complaining the item extruding is so slow :P 01:43 VanessaE kahrl: item extruding is on a per-texture basis, right? 01:43 kahrl per-item 01:44 VanessaE hrm. Even with nodeboxes, the extrude code has to be run? 01:44 kahrl yeah, at least with preload, it extrudes every item (including every node even if it can't ever be in the inventory) 01:44 VanessaE and item extruding IS slow - I mentioned once of a VERY simple test that proves it, and no one listened. 01:45 VanessaE well it seems to me that if you're extruding textures for nodes/nodeboxes you're already doing it wrong. 01:45 VanessaE that code shouldn't be run on anything that isn't flat, sprite, or plantlike 01:45 VanessaE (and even then, it should be made possible to just skip it) 01:46 kahrl well not extrude, but it has to generate a mesh for the item and do the render-to-texture magic 01:46 VanessaE but that's not the argument here. 01:46 VanessaE care to try my test? 01:47 kahrl I'm busy doing some other stuff but later I can 01:47 VanessaE open GIMP. Make a file, 512x512 pixels. give it an alpha channel. Fill it with some pattern that sets approximately 50% of the pixels to black. The other 50% transparent. 01:47 VanessaE save that as a png. Make say 10 copies of it. 01:47 kahrl oh, I know that's slow 01:47 VanessaE but give each a subtle difference so they have different hashes. 01:48 VanessaE you'll find those 10 images take longer than the entirety of vanessae_game combined (default textures) 01:48 VanessaE I discovered this last year 01:48 VanessaE and as usual no one listened to me. 01:49 kahrl it's more like everyone knows that the code is slow in many cases but no one has a good idea how to improve it 01:49 VanessaE then why complain about the number of nodes? let ME and MY USERS worry about whether it's slow or not :) 01:50 VanessaE you guys got bigger fish to fry, I'm sure 01:50 kahrl it's just that there shouldn't have to be a googolplex of variations of every node 01:51 VanessaE I can see your reasoning, but to be, there may very well have to be if you want the maximum possible amount of flexibility in those nodes 01:51 VanessaE to me* 01:52 Exio the way MC manages the stuff is having, for example, one node for all the wood and a "param1" for the type, param2 for the rotation (or "whatever" that can be translated to MT), and other for slabs, other for double slabs, and so on 01:52 Exio (for hmmmm) 01:53 kahrl MC has a separate .java file for every block (node) so they can do that 01:53 Exio MC doesn't have a mod api 01:53 VanessaE what you're talking about is so fundamental a change in the way nodes are defined that it basically means ripping up every mod out there that uses some kind of variation from node to node and rewriting them from the ground up. 01:53 Exio they can hardcode stuff and be happy forever 01:53 kahrl VanessaE: no I'm not 01:53 VanessaE exio ^^^ 01:53 Exio what 01:54 Exio yes, i know it isn't what MT is doing by now 01:54 VanessaE defining a node model and then assigning textures to it separately would require a completely different way than we use now 01:54 Exio and what is wrong with having a better way to do the same stuff? 01:54 VanessaE nothing at all 01:55 VanessaE except I'm trying to get this done NOW since the change is so easy to make,. and leave the more complicated stuff for another day 01:55 VanessaE -. 01:55 kahrl if you just change this the nodedef serialization becomes extremely inefficient 01:56 kahrl sure it works 01:57 Exio working stuff != stuff working in the right way 01:57 VanessaE I'm saying to put that $10 radiator hose in now because it'll get you to work, and worry later about the engine overhaul the car needs. 01:57 Exio a hacky-fix because it is easier-fix-to-do and will keep forever 01:58 Exio the forever-temporal-fix 01:58 kahrl Exio: exactly 01:58 VanessaE Exio: a hacky fix that can be undone as quickly as it was applied? 01:58 kahrl it can't be undone without completely breaking your server, for example 01:59 kahrl and client/server compatibility hell 01:59 VanessaE kahrl: break it how? 01:59 Exio don't you have >4096 nodes defined? 01:59 VanessaE (other than having to remove a mod no one's using yet) 01:59 kahrl it will be used once the change is in 01:59 VanessaE Exio: yes, and only because I'm helping kacey test a new mod based on one of mine. 02:00 Exio lets imagine now MT reduced the node "limit", your server now would crash like it died 02:00 Exio did * 02:00 VanessaE kahrl: what do you mean "it will be used"? 02:00 Exio that was quickly undone! 02:01 VanessaE Exio: in which case I'd uninstall whatever mods are used the least and alias the dead nodes to something sane 02:01 Exio VanessaE: people will use 4000 mods 02:01 Exio defining 20 nodes each 02:01 VanessaE sure. so let them. but I don't care about them right now - I care about the one server and two clients that have this change in place for testing and the fact that it *works* 02:02 VanessaE let someone use 50, or 5000 modes, I just don't care about that. 02:02 VanessaE they're not stupid and can decide for themselves what is and is not a good use of their time, memory, network bandwidth, etc. 02:02 Exio the idea is not add more hacky-fixes to minetest 02:02 Exio it already has a lot of them 02:03 Exio VanessaE: and you were the one who used the "I" 02:03 Exio you may not care 02:03 hmmmm indeed 02:03 VanessaE well whatever, I give up. 02:03 hmmmm having a higher limit just reduces creativity 02:03 VanessaE hmmmm: do you play? 02:03 hmmmm if we make the limit 65534, people will just use up that many 02:03 hmmmm and then they'll complain more 02:04 hmmmm "make the limit 4.2 billion" 02:04 hmmmm 32678 in all directions isn't enough 02:04 hmmmm gotta increase that too 02:04 Exio VanessaE: you are complaining about minetest being slow 02:04 Exio and then want it be more slower :P 02:04 hmmmm an increase in these limits would just encourage inefficiency 02:04 VanessaE *facepalm* I refuse to listen to slippery slope arguments. 02:04 Exio haha 02:04 hmmmm well 02:05 VanessaE well nothing. 02:05 hmmmm i guess you don't care for inductive inference 02:05 hmmmm which is a primary facet of logic 02:05 hmmmm we've seen it before 02:05 hmmmm i want to come up with something smarter instead of just adding more flack 02:05 VanessaE I don't care for claiming things will be stretched to their logical extremes and then using those as reasons not to do something. 02:06 VanessaE people will *always* stretch things to the extremes. 02:06 hmmmm right, like you :) 02:06 hmmmm stretched the node usage to the extreme 02:06 VanessaE hardly. 02:06 hmmmm and beyond 02:06 hmmmm hell, you're the one who got it higher than 2048 02:06 hmmmm 2048 is pretty extreme. i think i use 68 nodes total myself 02:07 Exio i really think a "200~400" nodes is pretty ok for a featured-gameplay 02:07 hmmmm now, i'm not saying we shouldn't increase the limit 02:07 VanessaE ok, server's back to the normal unmodified code and the extra mod removed. nothing broke. 02:07 hmmmm i'm just saying it's stupid to do so 02:08 hmmmm for reasons other than technical 02:09 hmmmm i hope the colorlike node type will help matters... but like you said, it won't help you much 02:09 hmmmm i'm sure we'll think of something. 02:09 Jordach hmmmm, so colorlike adds more to the node data usage 02:09 Jordach so why dont we have colored torches then 02:10 hmmmm erm, no it doesn't 02:10 kahrl VanessaE: out of curiosity, what is the mod that consumes the most nodes on your server? 02:12 VanessaE um 02:12 VanessaE I would guess pipeworks or moreblocks. 02:12 VanessaE (moreblocks w/stairsplus) 02:12 VanessaE colorlike is stupid, I'm sorry 02:13 VanessaE you proposed, when you last said, eight colors if the node has facedir. 02:13 Exio what do you suggest for that 02:13 hmmmm seriously, you say it's stupid, but those are the constraints 02:13 hmmmm if you can come up with something better, code it yourself! 02:14 hmmmm doing > whining 02:14 VanessaE those are the constraints because you're still trying to treat this like it's an old 8-bit system (i.e. limited disk/network) 02:14 Exio (without making MT 25x times slower) 02:14 hmmmm zz 02:14 hmmmm then fork minetest and make param2 bigger 02:14 * VanessaE sighs 02:14 hmmmm make param1 store colored light while you're at it 02:14 VanessaE point: missed. 02:15 hmmmm so we'll have param2 be 2 bytes, param1 will store 3 bytes, leaving 3 bytes for content 02:15 hmmmm there, 1.67 million nodes 02:15 Exio loading the world will take about other 1.67 million of seconds, no? 02:15 hmmmm haha :/ 02:15 hmmmm actually probably twice as long 02:15 VanessaE sure if you're stupid about how to use the damn map data 02:16 hmmmm no idea, actually it might take the same amount of time 02:16 VanessaE I mean come on, this stuff was figured out decades ago. Dynamic packet length. 02:16 hmmmm if most of that data is 0 it'd compress well 02:17 VanessaE you only need a single spare bit somewhere in the existing data (and as I recall, we have three). if that bit is set, then and only then does that node have extra data bytes to specify more parameters/features/whatever. If it's not set, use the format we use right now. 02:18 hmmmm then how do i access the node at (2, 5, 7) in the mapblock? 02:18 VanessaE by...breaking the map into proper database records? 02:18 hmmmm ??? 02:18 Jordach stop using sql, thats why 02:18 VanessaE tag it by whole chunks if you have to 02:18 Exio hmmmm: how do i access the node at (2, 5, 7) in a fast way* 02:18 Jordach levelDB is superior 02:18 VanessaE (tag = use that set/not set bit) 02:19 hmmmm exio: I am asking vanessa a rhetorical question so she realizes that it's the worst idea ever proposed 02:19 VanessaE *shakes head* 02:20 hmmmm zzzzzzzzzzzzzzzzzzzzzzzzzz 02:20 Exio hmmmm: i guessed :P 02:20 hmmmm if you guys can do better, DO IT! 02:20 hmmmm seriously! 02:20 hmmmm stop talking about it and actually do it! 02:20 VanessaE how many nodes do you suppose the average map has stored? a few billion on redcrab's server? 02:20 VanessaE (5 or 6GB map as I understand now) 02:20 Jordach VanessaE, its gone beyond 10gb 02:21 VanessaE that big eh? wow. 02:21 Jordach its either going to collapse from size soon 02:21 Jordach sql isnt meant for that level of usage 02:21 hmmmm lol 02:21 kahrl wut? 02:21 VanessaE how do you suppose enterprise companies store those massive databases they have? I'll give you a hint, it ain't with SQlite and it ain't without using some kind of easily-indexable records. 02:22 hmmmm erm 02:22 hmmmm in all seriousness 02:22 VanessaE I'm being serious. 02:22 hmmmm sql dbs are way bigger than you give it credit for 02:22 hmmmm you only get into NoSQL and what not at the google level 02:22 VanessaE I'm no coding noob and don't appreciate being talked down to like such. 02:22 VanessaE I just don't have (or care to gain) any skills with C++. 02:23 VanessaE I'm trying to do everything I can to help this project along, in any direction I am able to push it. 02:23 kahrl VanessaE: that was meant as a reply to sql isnt meant for that level of usage 02:23 VanessaE kahrl: I know. 02:23 Jordach kahrl, the map format likes to break when its used too much 02:24 kahrl sql != sqlite 02:24 hmmmm and he's gone. 02:25 hmmmm hey kahrl, have you seen that missing mapblock schematic bug any more after that? 02:25 kahrl I haven't tried 02:25 hmmmm mmmmeh 02:25 hmmmm well at least we're aware of it 02:27 VanessaE so, are you going to address my comment on extending the map format or just continue to treat me as an idiot? 02:28 hmmmm variable length packets... i shouldn't even have to point out the problem with this 02:28 hmmmm where's the random access 02:28 hmmmm we need that. 02:28 VanessaE hash/link table. 02:29 VanessaE Berkeley Softworks had that figured out 30 years ago for crying out loud. 02:29 hmmmm so you're saying we should have a 4096 element hashtable for every single mapblock that we use to look up the locations of mapnodes to modify for the purpose of........ what again? 02:29 kahrl and if you change the first node in a mapblock, you redo all links? 02:29 VanessaE we called it VLIR back in the day, I'm not sure what the proper term is now. 02:29 hmmmm anyway 02:30 hmmmm sounds like a great idea, you should fork minetest and try it out for yourself 02:30 hmmmm if it works out well, you have every right to tell us how much better vanessatest is. 02:30 VanessaE "you should fork"..."why don't you fork it".... *sigh* 02:30 hmmmm yae 02:31 hmmmm because doing is better than talking about 02:31 hmmmm 'sides, i already have a lot of work cut out for me to do 02:32 VanessaE well I give up. You won't find any better way either because you're too busy worrying about other stuff 02:32 VanessaE and no one else will bother because no one else even cares. 02:34 VanessaE this discussion started out as a request for temporary change that fixes a bug c55 himself shouldn't have existed, and morphed into a big long philosophical discussion about every other damn thing except what I original asked. 02:34 VanessaE +ly 02:35 kahrl it's not a bug 02:35 kahrl but whatever 02:35 VanessaE c55 said it was. 02:35 kahrl link? 02:35 VanessaE it was supposed to be limited to 65534. 02:35 VanessaE gimme a few to find it. 02:36 VanessaE looks like back around 0.4.1 02:37 VanessaE http://irc.minetest.ru/minetest-delta/2012-09-08#i_2490113 02:38 VanessaE "due to a leftover constant" = a bug, in my book 02:40 kahrl "by the way, if the size of that static table is set to 65536, it will always consume at least 50MB of RAM" 02:40 kahrl ^ if that is fixed I'm okay with upping the constant 02:41 VanessaE kahrl: which is why I said up it anyway and fix the reason for the 50MB-usage later. 02:41 Exio fixing things later = fixing it in 4 years 02:42 Exio (or never, what works better) 02:43 VanessaE Exio: and what is there to force it to get fixed *at all* then? 02:44 VanessaE (in other words, it's just as likely to go unchanged for 4-years-to-forever if left in its current state as it would be if it were tweaked as previously indicated. 02:44 VanessaE (and I'm not trying to imply that this is what will happen, but considering the date on that log entry...well...you do the math) 02:55 hmmmm [07:42 PM] first of all in order to do that we'd need to change it from a statically sized array to a vector 02:55 hmmmm [07:42 PM] and then you also need to fix node id mapping serialization 02:55 hmmmm [07:43 PM] it's not a one liner fix 02:56 VanessaE hmmmm: forget about it. you guys have already made up your minds what will be done and when. 02:57 kahrl it's not a one liner but not that tough either 02:57 kahrl there's only about a dozen references to MAX_CONTENT in the codebase 02:58 kahrl most of them trivial to change to a dynamic sized vector 02:58 kahrl the only slightly tough one is the serialization 03:03 kahrl is there any current pull request that would update the protocol version? 03:13 Tesseract kahrl: Sapier had a animation one that could be done in a nicer way with a protocol version increment. 03:28 kahrl #774? that doesn't increase the protocol version 03:29 Tesseract kahrl: I know. He said it could be done in a better way but he didn't want to raise the protocol version for it IIRC. 04:43 hmmmm oh pilzadam 04:43 hmmmm for your most recent commit, you should've passed a reference to the player name, not the entire string itsel 05:40 hmmmm when a lua api returns an array to you, do you like it to start at 0, or 1? 07:25 proller try #7: ipv6. 07:25 proller 5 minutes to merge.. 07:28 proller 3.. 07:32 proller 0! 07:37 Calinou :cyisfor happy face: 07:45 celeron55 hmmmm: of course at 1; lua's built-in array handling stuff expects that 07:45 celeron55 you don't want those to break 07:49 celeron55 (also all users expect that because it's how it's designed9 07:49 celeron55 )* 08:06 celeron55 also, it's funny how VanessaE alternately complains about slowness and proposes things that would make everything so slow it wouldn't be even funny 08:06 celeron55 8D 08:15 Calinou propose what :P 11:59 proller https://github.com/minetest/minetest/pull/659 11:59 proller no json in config 8( 12:07 Calinou json, overrated since 2009 12:10 sapier json is very very ugly if you need to pass it to lua 12:11 sapier e.g. the online server list 12:18 proller Calinou, why? 12:18 proller sapier, lua more ugliest 12:20 sapier even in c++ parsing json isn't quite beautiful ;-) 12:20 proller show me how to easy get from lua structure like {"distance":0.01, "center":{"x":5,"y":-100,"z":42},} 12:21 PilzAdam {distance=0.01, center={x=5,y=-100,z=42}} 12:21 proller PilzAdam, how to transfer in in c++ 12:21 proller in json: params["center"]["x"].asFloat() 12:22 proller for getting center.x 12:22 proller very easy 12:22 sapier it's not meant to be transfered proller lua isn't a datastructure but a full language 12:22 sapier while json is a container format only 12:22 proller for math mapgen needed only container 12:23 proller language useless 12:23 sapier if it's not meant to be changed make it hardcoded ;-) if it's meant to be changed there should be a reasonable way to add a gui by some time 12:24 proller maybe 12:24 proller but there is 10-20 params depend on generator 12:24 proller its complex gui 12:25 sapier of course in best case it'd be a "advanced settings" gui hidden by default 12:25 proller for stipid sphere: radius, center(xyz), offset(xyz) 12:26 proller i can commit it, and then think how to better handle params 12:27 kahrl one problem is that settings.h is included almost everywhere 12:28 kahrl and now everything will therefore have to include json.h --> compile time nightmare 12:29 proller simplest is making lot of math_generator, math_center_x, ........ params, but they must be exist in 3 places 12:29 sapier my concern is more that we had to implement/include json in lua if there was a setting in json format 12:29 sapier do all those parameters really need to be modified? 12:29 proller by default all ok 12:30 proller but you can want to change 3-5 params 12:30 sapier no my question is does it really make sense to change single parameters? 12:30 proller yes 12:30 proller you can change world 12:31 sapier does anyone except those who designed it understand meaning of parameters? 12:31 proller make very different from default 12:31 proller everyone who know someting about fractals 12:31 sapier I guess ppl want to change e.g "0-100%" below water level mountain height ... 12:32 sapier if I remember correctly I wouldn't have any idea what parameter to change to get what I want 12:32 sapier or am I completely wrong? 12:33 proller params can be described 12:34 proller for mandelsponge you can want to change pillars weight 12:34 proller or level of sponge 12:34 sapier I don't have any Idea what you're talking about ;-) 12:35 proller its hard without trying 12:36 sapier are we talking about a very special mapgen not to be used by many ppl? 12:36 proller http://en.wikipedia.org/wiki/Menger_sponge 12:36 proller sapier, yes 12:37 proller its more for fun 12:37 sapier ok don't get me wrong I'm absolutlely positive about different mapgens in minetest, but do you expect anyone to even try it without a easy gui? 12:38 proller 3-5 peoples - yes 8) 12:38 sapier and moreover I guess ppl would want to change the settings per world not globaly 12:38 sapier so what about hardcoding reasonable default values giving ppl a gui to choose on world creation and maybe save those settings as templates to separate files? 12:39 proller but its impossible now via gui 12:39 sapier proller that's why I added formspec mainmenu .. you can add a button to show your own gui 12:40 proller i try it when it would be merged 12:40 sapier ok saving json from formspec/lua would be difficult but at least minetest.conf wouldn't explode by parameters most ppl don't even need 12:40 kahrl sapier: these are mapgen params so they will be saved in map_meta.txt 12:41 proller btw it was saved it map_meta.txt in json string, but i remove it because nobody wants json in config 12:41 sapier yes for that world of course but maybe someone want's to save setting to reuse it in other worlds 12:42 proller now in my branch all params is hardcoded on good default, you spawn inside mandelbox and can explore it 12:42 proller if someone wants to change params - he must recompile 12:43 proller next step - trying to make gui 12:43 proller but separate from mapgen 12:43 sapier proller you should do it in mainmenu branch it's formspec support is improoved drasticaly 12:44 proller ok 12:44 proller but firs i want to commit hardcoded into master 12:45 proller https://github.com/minetest/minetest/pull/659/files 12:45 sapier is there a way to decide which mapgen to use on world creation? 12:45 proller in config 12:46 sapier :-/ I think it'd be way more usefull if it was a parameter 12:46 proller mg_name = math 12:46 sapier maybe I should add dropdown support to formspec too 12:46 proller someone planned make gui for mapgen select 12:46 PilzAdam we need a gui to set all mapgen params at world creation, using the settings in minetest.conf only as default value 12:47 sapier PilzaAdam yes but imho that gui should only be shown if user pushes a button 12:47 sapier no need to bother users with parameters they most likely don't understand 13:23 sapier hmmmmm I guess the voxel manipulator already contains lots of parts for transactional support at leat it's concept is already quite similar 13:58 sapier can someone explain to me what exactly conlock is expected to protect? 13:59 proller and final! https://github.com/minetest/minetest/pull/659/files 14:00 proller if no BIG objections - will merge 14:00 PilzAdam ummmm 14:00 PilzAdam you have merged upstream/master into your working branch 14:00 PilzAdam thats the worst thing you can do with git 14:00 proller i will squash 14:00 proller git worst itself 14:01 PilzAdam remove the merge commits and rebase instead 14:01 PilzAdam no, git is awesome, you just have to know how to use it 14:01 proller i will commit to master with squash 14:01 proller and i hate rebase 14:01 sapier git is awsome because of not limiting anything ... some ppl call it horrible because of exactly same thing ;-) 14:03 proller unlimited ways to get conflict or lost commit 14:04 sapier git is designed for extreme concurrency and (version control) skilled programmers 14:09 sapier https://github.com/minetest/minetest/pull/785 just a small fix plz pull 14:11 sapier yes it's very unlikely to happen but in very rare cases (most likely due to other issues) this happens 14:12 PilzAdam yay! another useless mapgenerator 14:12 proller yes! 14:12 sapier I'm gonna add a dropdown to select mapgen in mainmenu 14:13 sapier more detailed settings is up to the one responsible for that special mapgen ;-P 14:14 PilzAdam for your most recent commit, you should've passed a reference to the player name, not the entire string itsel 14:14 PilzAdam ^ hm? 14:14 hmmmm yeah, i really don't see a reason why you'd do an entire copy just to print out a name 14:16 PilzAdam hmmmm, btw, I get "warning: multi-character character constant [-Wmultichar]" in mapgen.cpp:692 and 688 14:17 hmmmm yeah, i know. 14:17 hmmmm it's a stupid warning, really. 14:18 sapier ? do we use utf16? 14:18 kahrl does the standard define bit order of these? 14:20 kahrl proller: // Mapgen Singlenode parameter read/write 14:20 kahrl should be Math 14:20 PilzAdam "proller authored 13 minutes ago" -> dont rewrite history 14:21 kahrl new commit then 14:21 proller ok 14:22 hmmmm the rule is a new commit is needed after 5 minutes and nobody seems to have pulled 14:22 PilzAdam hmmmm, nope, 10 minutes 14:22 kahrl time is relative 14:23 PilzAdam well, 5 minutes if you are moving really fast compared to the GitHub server 14:23 hmmmm lol 14:23 hmmmm new rule: nobody can commit anything if they're moving relative to github near the speed of light 14:33 proller and can commit in past, if can moving faster than light 14:43 hmmmm anyway last night i layed down some initial code for LuaVoxelManip (that's the name i decided on) 14:44 hmmmm i've decided that it's not going to be overcomplicated, and pretty close to the bare metal 14:44 sapier ok I'm looking forward to it 14:45 hmmmm as a demonstration i'll make a simple complete lua mapgen 14:45 Jordach paramat will be happy 14:45 Jordach i suggested that a while back 14:47 proller hmmmm, 10x slower& 14:48 hmmmm i'm going to do a side-by-side comparison of the same thing done in the core and in lua using pure get/set_node calls 14:48 hmmmm i suspect it'll be reasonably good though 14:50 sapier hmmmm do you intend to support async api? 14:50 hmmmm i can't with this 14:51 sapier for what reason? 14:51 hmmmm because it hardly spends any time at all in the core 14:52 sapier meaning everything is done in lua? 14:53 hmmmm yes 14:53 Exio wait, a random thing 14:53 sapier but that's exactly what async is intended to improove? 14:53 Exio there was something about making the light_update cause after the on_generate, instead of "after every set_node" 14:54 hmmmm there's an update lighting function 14:54 Exio what's up with that? for making old-set_nodes faster + new api that is even faster 14:54 hmmmm and you can also cancel the mapgen's calcLighting) 14:54 hmmmm so no wasted effort here 14:55 sapier exio of course if there's an option to improve old one 14:55 hmmmm someone's going to have to rewrite VoxelArea in lua 14:55 hmmmm ugh 14:55 hmmmm btw i hate lua 14:56 proller +1 14:56 proller lets make js api 14:57 hmmmm lol 14:57 celeron55 as if hmmmm liked javascript 8D 14:57 hmmmm i wrote a cloud fractal generator in javascript 14:57 hmmmm the pixels were CSS rectangles of different colors 14:58 sapier hey after scriptapi split you can add almost any backend ;-P it's just a lot of work 14:58 celeron55 javascript is like lua designed in a haphazard ad-hoc manner 14:58 sapier you only have to preserve cpp_api interface 14:59 sapier imho lua isn't worse than any other lightweight script language 14:59 celeron55 lua is pretty much the best all-around embeddable scripting language 14:59 sapier it lacks object orientation support but you can get close to it 15:00 celeron55 some other language might be slightly better as a language, but their implementations just plain suck 15:00 hmmmm wondering, how do you set members of objects to values in lua? do you have to muck around with that metatable crap? 15:00 sapier and once you have understood stack concept even interfacing from c/c++ isn't that bad 15:01 sapier you need to set up your stack correctly and then push the new value 15:01 celeron55 hmmmm: uhm 15:01 sapier at least if you mean objects as tables 15:02 celeron55 hmmmm: that question is somewhat arbitrary because lua doesn't really have a thing called "object"; the definition of that term is really specific to how you use lua 15:02 hmmmm right, right, i know, everything's a table 15:02 sapier get table index 15:02 celeron55 but the answer probably is no, you just say myobject.foo = 1 15:02 sapier get field index 15:02 sapier push new value 15:02 celeron55 or if you mean in C, just set it as in a table 15:03 celeron55 hmmmm: you use metatables in lua in order to do stuff like the object prototype pattern that javascript uses 15:04 celeron55 or mechanism 15:04 celeron55 whatever 15:04 celeron55 in lua it's a pattern, in js it's a mechanism 8) 15:05 sapier hmm I thought I had some example code in lua mainmenu but there's only code to read elements of a table 15:06 sapier maybe writing s similar 15:06 hmmmm i'm not going to bother with it, i just have multiple returns 15:07 sapier wait you only want to return it? 15:07 sapier that's quite easy 15:07 hmmmm eh 15:07 hmmmm i'm going with the easiest route here 15:07 hmmmm don't really feel like doing anything fancy 15:08 sapier if you want some return value of style 15:08 sapier { 15:08 sapier somename1 = val1, 15:08 sapier somename2 = val2, 15:08 sapier } 15:08 sapier that's not even fancy :-) 15:09 hmmmm local vm = minetest.env:get_voxelmanip(); local ndata, emerged_p1, emerged_p2 = vm:read_chunk(p1, p2); vm:write_back(); vm:update_map(); 15:09 hmmmm of course, s/;/\n/ 15:10 hmmmm vm:write_back(ndata), rather 15:10 hmmmm does that seem sane to you guys? 15:11 celeron55 doesn't really matter as long as you're open about someone possibly reworking the returns after seeing the whole thing 15:11 PilzAdam hmmmm, you can use ; instead of \n in Lua AFAIK 15:11 hmmmm yeah i really don't care, you guys can do whatever you'd like 15:11 hmmmm pilzadam: good to know, thanks 15:12 PilzAdam but its against our code style guidelines in Minetest 15:12 celeron55 it's discouraged practice because it makes code less predictable to read 15:12 sapier hehe hmmmm and his semicolons ... a fantastic tale in 3 books :-P 15:12 celeron55 but it makes single-line examples like that valid without explanations 15:13 sapier "less practical to read"? guys I just copied that code added \n to understand it ;-) 15:22 PilzAdam proller, why isnt "sphere" solid? 15:23 proller disable dungeons 15:23 proller and caves 15:23 proller in config 15:24 proller it uses v7 code for everything 15:25 PilzAdam still not solid 15:27 proller maybe cave disabling ignored.. 15:27 proller will check.. 15:29 PilzAdam http://mg.viewskew.com/mgoblin_media/media_entries/31/Unbenannt.png 15:29 PilzAdam ^ I doubt this is caused by the cave generator 15:30 proller woow 15:31 proller looks like "hello" from v7 15:31 proller show config about mapgen 15:31 proller and water level 15:33 PilzAdam http://mg.viewskew.com/u/pilzadam/m/math-sphere-bug/ 15:33 PilzAdam see description 15:35 proller ok will fix in ~ hour 15:36 PilzAdam have you pushed that without testing? 15:46 sapier how is singlenode mapgen intended to behave? 15:46 PilzAdam sapier, it places "mapgen_singlenode" everywhere; if its not defined, then air 15:46 Calinou ^ 15:46 proller fuck yeah: 19:46:36: ACTION[ServerThread]: proller [::ffff:192.168.0.241] joins game. List of players: proller 15:47 proller PilzAdam, it tested it 1-2 months 15:47 sapier ok so if not defined it's an empty map ... exactly what I experience 15:48 PilzAdam proller, I always test it again before pushing to upstream 15:48 proller PilzAdam, its from lower water 15:50 proller and you have sphere, maybe not complete ... 8) 15:52 sapier guys it's great you always modify guiFormspecMenu.cpp/h in core 15:56 proller btw: Segmentation fault: 11 always when exiting in menu from game, freebsd 15:57 proller PilzAdam, its not bug, its mouse eat cheese 16:07 PilzAdam proller, also the lighting is not correct 16:15 celeron55 i don't think proller is going to be able to fix the lighting (it's a long-standing corner case in the chunk-based map generator) 16:15 celeron55 (most likely) 16:16 proller able, just disable calcLighting 16:16 proller and make nodes with max light 16:16 proller 60% complete.. 16:17 celeron55 what i mean is the one where the map generator isn't able to correctly handle a situation where the bottom part of something generates first and then the upper chunk is generated, in which case sunlight that incorrectly went into the bottom part isn't erased 16:18 celeron55 (0.3.x is able to handle it, IIRC) 16:19 PilzAdam proller, yea, fullbright would be fine I guess 16:20 proller its was working, but then i want trees and caves and other stuff and switch to v7 base 16:23 proller works! 16:25 proller https://github.com/proller/minetest/commit/8c37364fe445818197166302d6a4db607c5027f9 16:26 proller good? 16:30 proller commiting! 16:30 Calinou https://github.com/minetest/minetest_game/pull/181 16:33 sapier calinou I can see it's different but imho it's no change in quality 16:37 PilzAdam proller, wait, we dont change several things in one commit 16:37 proller oops 16:37 proller sorry 16:37 sapier Ooops ... PilzAdam I won't split the formspec mainmenu patch ;-) 16:38 proller call it "one Improvement thing" 16:38 PilzAdam sapier, the Lua menu is one thing 16:39 thexyz https://github.com/minetest/minetest/issues/784 16:39 sapier not exactly it contains online modstore offline zip support, gamemanager, modmanager ... ;-) 16:39 thexyz sapier: all in one commit? 16:39 sapier of course ... no way to split it reasonably 16:40 sapier at least not without adding mainenance nightmare of merging forward backward between different branches 16:41 sapier and as this pull request is delayed everytime it's growing even more 16:41 celeron55 Calinou: that's not better, it's just different 16:42 Calinou mostly 16:42 celeron55 the only way to make grass better without making it boringly non-textured is to implement pseudorandom selection of texture variations for nodes 16:42 Calinou ingame it tend sto look worse 16:42 celeron55 (not sure if there is an actual term for that; probably is) 16:42 Calinou "random textures" --sp614x 16:43 Calinou optifine (minecraft mod) lets texture makers use random textures, connected textures, repeating textures (that span on several nodes) 16:43 Calinou and there's a "natural textures" option which rotates textures randomly 16:43 sapier proller I just added mapgen selector to formspec mainmenu I guess this will allow a lot more ppl to test it 16:44 sapier at least IF it is added by some time 16:44 proller sapier, cool 16:45 proller intresting thing: in not random word you can move spawn point, or move world around spawn poing 16:48 sapier I wonder how the mapgen details page should be implemented 16:49 proller lot of param = value 16:49 proller with various types of values 16:53 sapier http://imgur.com/BXDdD8M I thought mor about the api to do that. yes there could be a single hardcoded page but I'd prefere to select it by mapgen name 16:53 sapier a button right of selector "settings" would open that mapgen settings page 17:23 hmmmm ohh sapier, i was going to do that 17:23 hmmmm once the formspec menu was finished 17:24 sapier hmmmm it's as finished as it ever will become without more testing 17:25 proller hmmmm, in v7 trees seed planed via boimes ? 17:26 hmmmm yeah 17:32 proller ridges too buggy 17:58 hmmmm 'too buggy' 17:58 sapier https://github.com/minetest/minetest/pull/774 changed set_animation to support client prediction of animation framerate ... problem with changing packets is breakage of compatibility 17:58 hmmmm you modified mapgen v7 i see.... 17:59 hmmmm the lighting member should be a mapgen flag 17:59 hmmmm and i was planning to put that in withe luavoxelmanip anyway 18:00 hmmmm the 'ridges' variable is not needed at all... you can set scale to 0 18:01 proller it was fast fix 18:01 hmmmm besides, if you merely disable 'ridges' like that, you screw up getGroundLevelAtPoint() 18:01 hmmmm yeah, we don't like fast fixes though.... 18:01 hmmmm same reason why we didn't simply change MAX_CONTENT yesterday 18:01 hmmmm doing things the proper way > doing things fast 18:02 hmmmm also holy crap line socket.cpp line 189 18:03 hmmmm you should really break up your lines to be around 80-90 if you're modifying files that aren't yours 18:03 sapier https://github.com/minetest/minetest/pull/693 what about merging this? it'll improve entity riding 18:04 Mallot1 hihihihi goood morning :D 18:04 hmmmm lol 18:04 hmmmm i don't even know what stepheight is 18:04 sapier it's a parameter used to decide if player can step up without jumping 18:04 hmmmm aah 18:05 sapier for entities this is hardcoded to ß 18:05 sapier 0 18:05 hmmmm this is for mobs to move around i guess? 18:05 hmmmm move around better 18:05 sapier mobs too yes but most benefit is if you ride a ostrich you don't have to jup on stairs 18:05 hmmmm reminds me of the new minecraft horse feature 18:05 hmmmm wow, we really outdid minecraft.... 18:05 sapier NO i've done this first! 18:05 hmmmm good job :)! 18:06 PilzAdam animals mod for MC was first :-p 18:06 sapier what's animals mod for mc? 18:07 sapier did they even steal my old name? 18:07 PilzAdam its "more creatures" or so 18:07 hmmmm moar oars 18:07 hmmmm moar biomes 18:07 hmmmm moar creatures 18:07 hmmmm moar mountain dew 18:07 sapier (-moar + more)*4 18:08 hmmmm whenever i read more ores i think of it said as "moar' 18:08 sapier :-) 18:08 sapier hmm reading the code again do we need a protocol change for that? 18:08 Jordach hmmmm, mountain dew sucks... 18:09 hmmmm well, more biomes is actually extra biomes, thank god they didn't call it "more", i think i'd want to shoot something then 18:09 hmmmm sapier, no it doesn't 18:09 Jordach BonusBiomes? 18:09 Jordach BiomesDX? 18:09 sapier so adding the stepheight isn't a big deal am I right? 18:10 hmmmm mod naming rule - if more than one word, they must either rhyme or share the first consonant sound 18:10 hmmmm sapier, it's not 18:10 PilzAdam hmmmm, nope, "Simple Mobs" 18:10 Mallot1 fdsgregfet4gtehthrry yay minetest!!!! 18:10 sapier rename to Multiple Mobs 18:10 hmmmm you only need to change protocol versions if you change something, or add it into the middle of the contentfeatures 18:10 Mallot1 i love ur imple mobs 18:11 hmmmm lol 18:11 sapier too late mallot1 ;-) 18:14 sapier is travis build broken atm? 18:15 PilzAdam seems so 18:16 hmmmm hmmmmmmmmmmmm 18:16 PilzAdam wait, no, the interface just sucks 18:16 PilzAdam sapier, just your branch is broken 18:16 hmmmm i need to make non-mapgen variants of updateLiquids and calcLighting really bad 18:17 sapier where to get information about what is broken? 18:18 PilzAdam https://travis-ci.org/minetest/minetest/jobs/8364855#L212 18:20 sapier where to find this in new interface? 18:21 whirm hi 18:21 whirm I'm getting this segfault when exiting to menu: http://paste.debian.net/12152/ 18:21 whirm on current master's HEAD 18:22 * hmmmm claps 18:22 hmmmm great job sapier 18:22 Exio disable the atlas? 18:22 sapier what are you talking about hmmmm? did to many things the last days ;-) 18:22 hmmmm your texturesource cleanup is still broken 18:22 hmmmm evidently 18:22 Exio hmmmm: he is using the texture_atlas 18:23 hmmmm exio, how do you know that? 18:23 Exio 20:19:10: ERROR[main]: TextureSource::buildMainAtlas(): Atlas is full, not adding more textures. 18:23 hmmmm ahh 18:23 hmmmm why would anybody be using the atlas exactly 18:23 Exio old config maybe 18:23 sapier maybe we should just ignore that parameter 18:23 whirm so, should I disable texture atlas? 18:23 hmmmm this is seriously all the more reason why we should rip it out by now 18:23 hmmmm whirm, yeah 18:23 whirm hmmmm: thanks :) 18:24 hmmmm sapier, instead of fixing that, would you rather remove the texture atlas entirely? 18:24 Exio wasn't the idea a bit ago, remove it completely? 18:24 PilzAdam sapier, I linked directly to the error line 18:24 hmmmm yeah but that didn't happen 18:24 sapier I thought I has a pull request somewhere removing it ... maybe I removed the pull request itself 18:25 hmmmm you must've because i don't recall seeing it when all that went on 18:25 Exio iirc there was one 18:25 sapier it's more than one error pilzadam I obviously was a little bit tooo careless when pushing it 18:25 hmmmm haha, pushing things to upstream without even making sure it builds? 18:26 whirm Ok, that worked :D 18:26 hmmmm don't start becoming like me 18:26 sapier seems to :-( shame on me :-/ 18:26 Exio now we have two hmmmms! :D 18:26 hmmmm oh wait you didn't push to upstream, just to your branch 18:27 sapier yes I can't push to master ;-) 18:27 sapier it's an experimental feature I didn't expect it to be merged as is 18:27 sapier still it should at least build :-) 18:30 sapier still where to find the travis information from pull window? 18:31 PilzAdam directly under the description or above the "merge" button 18:31 PilzAdam the link is called "Details" 18:32 sapier I don't have a merge button nor do I have a details link (at least I don't see one) 18:34 PilzAdam maybe there is no build for your pull request 18:35 sapier main menu? 18:35 PilzAdam (e.g. there is none for #785) 18:36 sapier ahh there "good to merge" ... ok no build for mainmenu 18:37 PilzAdam here is an outdated one: https://travis-ci.org/minetest/minetest/builds/8363176 18:37 sapier I always wonder why clang is that much faster 18:39 sapier ok I guess travis doesn't like me anymore :-) 18:42 celeron55 gcc is a convoluted chunk of premature output optimization 8) 18:43 PilzAdam anyone objects to removing builtin/item.lua:314 18:43 PilzAdam +? 18:47 hmmmm sapier 18:47 sapier what have I done? 18:47 hmmmm i put the new file in src/script/lua_api/CMakeLists.txt 18:47 hmmmm not getting compiled 18:47 hmmmm what sort of voodoo do i need 18:48 sapier what feature are you talking about? 18:48 hmmmm i am adding scripts/lua_api/l_vmanip.cpp 18:48 hmmmm trying to anyway 18:48 hmmmm i didn't need to delete the cmake cache before ever 18:48 hmmmm is that the problem? 18:49 sapier I don't think so 18:49 sapier but maybe rerun cmake 18:49 Mallot1 sorry 18:50 hmmmm whoops 18:50 hmmmm sorry 18:50 hmmmm i added l_vmanip.cpp to the one in src on accident 18:50 hmmmm and it failed to find it so it just executed the executable that was last there 18:50 sapier Mallot1 you shouldn't paste random chars here and chitchat goes to #minetest 18:50 hmmmm or... no i didn't 18:51 sapier you have to add it in lua_api directly 18:52 hmmmm alright i don't see what i'm doing wrong 18:52 sapier is there an error message? 18:53 hmmmm when i run cmake i get this "cannot find source file: $(CMAKE_CURRENT_SOURCE_DIR)/l_vmanip.cpp 18:53 hmmmm cmake error at src/CMakeLists.txt:381 (add_executable) 18:53 sapier wrong brackets 18:53 hmmmm o 18:53 sapier its {} 18:53 hmmmm well, i said i couldn't *see* what was wrong 18:53 hmmmm heh 18:54 sapier :-) yes they're quite similar 19:02 hmmmm what's the reasoning behind the name STACK_TO_SERVER(L)? 19:03 hmmmm oh nevermind 19:03 hmmmm why is it that you call a lua state a "stack"? it's way more than a stack 19:03 whirm ok, now I get this stacktrace while playing: http://paste.debian.net/12164/ 19:04 hmmmm ahh 19:04 hmmmm that must be the division by 0 19:04 hmmmm whrim, that's a rare occurance, just continue on playing, we'll fix it though 19:04 celeron55 i've never seen a SIGFPE that wasn't a division by zero 19:04 whirm I had it twice in 2 minutes 19:05 hmmmm interesting 19:05 celeron55 do you use the default chunk size? 19:05 hmmmm celeron, it's something that proller kept pointing out but he was never able to get a backtrace or anything since it wasn't consistent 19:05 whirm celeron55: I think so, let me check 19:06 celeron55 "chunksize" setting 19:06 hmmmm it has to do with proller's extra large caves 19:06 hmmmm he's using indev 19:07 whirm celeron55: yup, default chunksize, those are my settings: http://paste.debian.net/12168/ 19:07 proller again huge caves 19:08 whirm is there anything I can do to help debugging it? 19:08 hmmmm not really, we know exactly what the problem is, thanks though 19:09 proller (float)(ps->next() % ar.Z) + 0.5 19:10 sapier hmmmm yes stack isn't quite exact but most of time you only do stack operations on the state so I thought stack is more fitting 19:10 hmmmm ar.Z is 0, which means max_spread_amount - max_tunnel_diameter / 2 = 10 19:11 hmmmm max_spread_amount = 16, 16 - 12 / 2 = 10, which means max_tunnel_diameter was 12 at that time, which is indeed a possibility 19:12 hmmmm trying to decide whose fault this is 19:12 hmmmm nobody could've known that it crashes if the planets align 19:12 hmmmm so it's celeron's fault who originally wrote that cave code 19:12 hmmmm therefore the fix belongs in makeCave itself 19:13 hmmmm s16 more = MYMAX(max_spread_amount - max_tunnel_diameter / 2 - insure, 1); 19:14 proller seems good 19:15 sapier 2 - insure ? hope insure never gets 2 19:15 hmmmm insure is a constant 10 19:15 hmmmm besides, order of operations 19:15 hmmmm so that's irrelevant 19:15 sapier yes realized that :-) 19:17 whirm also, the farmesh thingie is broken too, right? (I'm messing with all the settings :D) 19:17 hmmmm farmesh is completely broken 19:17 Exio haha 19:17 hmmmm anyway whirm, download the newest version of minetest, should fix your bug 19:17 * whirm pulling 19:21 proller and i can set max huge cave to 200 19:23 whirm seems to work, thanks 19:23 celeron55 hmmmm: the cave code that is written by me sure can crash if it's stretched to do stuff it wasn't meant to do 19:25 celeron55 it's operation is based on certain things fitting inside each other that may not necessarily do so with some kind of input 19:25 sapier oooops: 19:25 sapier /home/sapier/minetest.folder/minetest_core_contributions/minetest/src/mapblock_mesh.cpp:682: TileSpec getNodeTile(MapNode, v3s16, v3s16, MeshMakeData*): Assertion 'facedir <= 23' failed. 19:26 sapier just standing around 19:26 celeron55 why is that an assertion 19:27 celeron55 there's nothing to actually protect that value from being <= 23 because it's value is context-dependent 19:27 celeron55 meaning* 19:27 celeron55 >23* 19:27 * celeron55 tries to write too fast 19:33 celeron55 i think it should assume facedir=0 if it isn't a sane value 19:49 roptat hello 19:49 roptat I just wanted to say thank you 19:50 roptat for all your work 19:50 roptat I really enjoy minetest 19:51 Jordach roptat, -> /join #minetest for this 19:51 roptat Well, I also have a bug to report 19:51 roptat but can't find your bug tracker 19:52 Jordach https://github.com/minetest/minetest/issues 19:52 Jordach ^ there 19:52 roptat ok thanks 19:59 celeron55 has somebody looked into this? http://forum.minetest.net/viewtopic.php?pid=94140#p94140 http://forum.minetest.net/viewtopic.php?id=6225 20:00 celeron55 has someone in here tried to go on that server and see if the problem can be reproduced? 20:00 celeron55 if not, someone should take that as their priority 20:03 celeron55 that really shouldn't happen on any working combination of server and client 20:14 Taoki hmmmm: Since you're working on the voxel manipulator, I thought those screenshots would motivate you: http://forum.minetest.net/viewtopic.php?pid=96297#p96297 Currently my system uses set_node, but once you're done with that and it's upstream I will switch to it, which should be a TON faster 20:15 Taoki Well, granted I can use the voxel manipulator to send a list of nodes to the engine, under the form { {pos1, node1}, {pos2, node2} } 20:16 Taoki http://i43.tinypic.com/2vx0ep1.png 20:17 Taoki Imagine when we can have towns +twice larger than that, also containing way bigger buildings, with nearly no resource efford at all :) 20:40 kahrl what's up with tile.h:193? 20:41 sapier1 guess there was something lost 20:43 sapier1 nope no define missing in code 20:43 sapier1 maybe someone wanted to add a new one and was interrupted forgetting about it later 21:04 proller falling liquid works! 21:04 proller make it optional ? 21:15 proller how to call lua nodeupdate from c++ ? 21:16 sapier1 is nodeupdate member of minetest? 21:18 sapier1 if it is getfield name of function 21:18 sapier1 push all parameters 21:18 sapier1 do a pcall 21:18 sapier1 fetch results 21:20 celeron55 oh god nodeupdate... once again a lesson in the topic "anything you release in an unstable version will actually be final, no matter what you think" 21:20 sapier1 hehe have a look at german constitution it's a draft for more than 50 years now ;-) 21:21 celeron55 8D 21:21 proller its in builtin 21:21 sapier1 so a local variable? 21:21 proller celeron55, its.... too hard to.. change.... 21:21 celeron55 it's a global variable probably 21:22 celeron55 thus in C it's a field in the globals table, however it's accessed (can't remember) 21:22 sapier1 then it's even more easy but you shouldn't rely on builtin features from c 21:22 proller function nodeupdate(p) 21:22 sapier1 yes global 21:22 celeron55 sapier1: it's allowable to rely on them 21:22 proller its need to make real waterfalls 21:22 sapier1 I wouldn't do so as it could be set to nil by anyone 21:22 celeron55 you can break everything horribly by messing up in lua anyway 21:23 celeron55 that's a completely useless concern once again by you 21:23 sapier1 yes but I'd check it at least in core 21:23 proller show any lua function called from c please 21:23 proller and i will copypaste 8) 21:23 sapier1 have a look at mainmenu branch it's done there multiple times 21:23 celeron55 of course it should be checked, but it's not like it couldn't be done 21:23 proller ok 21:24 sapier1 if there's no way to avoid it of course ... in mainmenu it's an explicit interface for example 21:24 celeron55 well it's a fact that there's no C++ implementation of falling nodes, it's all in builtin 21:25 sapier1 ok ok :-) 23:12 proller very hard to call lua func from map.cpp 23:12 proller no api vars 23:14 proller no server no emerge nothing