Time Nick Message 00:02 acerspyro Kodexky: ? 00:03 acerspyro Did you mean ctcp acerspyro version? 00:03 acerspyro oh 00:03 acerspyro I see what it does :P 00:03 Kodexky sorry faulty mouse. once i buy a new one i'll burn this one. 00:04 acerspyro lol 00:04 acerspyro Just buy one already! 00:05 acerspyro Can't be that broke that you have to wait to buy yourself a new mouse! 00:06 Kodexky not broke, just chained to the keyboard trying to avoid a deadline. 01:53 VanessaE those broken mapblocks can actually look like artsy at times: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2001122015%20-%2008%3a49%3a29%20PM.png 01:53 VanessaE look kinda artsy* 02:07 hmmmm that's because those colored blocks use up so gosh darn much of the content-space 02:07 hmmmm the probability is high it'd be one of those types of blocks 02:09 VanessaE yep, but it looks neat though :) 02:37 paramat that looks good 02:51 paramat http://cubeupload.com/im/rZWVwm.png lordawe has posted weird mapblocks too, note moretrees is in use in most of these screenshots, perhaps a connection? 03:03 VanessaE paramat: the ones I was posting are caused by the spawn_l_tree call in the engine 03:04 paramat ah that 03:04 VanessaE it pre-allocates space for a tree, but sometimes exceeds it before the map can be fully generated, and the result is garbage mapblocks 03:05 paramat can't see a fixing commit.. 03:05 Zeno` Still working on it 03:05 paramat cool =) 03:05 hmmmm that's the theory, at least 03:05 Zeno` it's the theory, yes :) 03:05 hmmmm in practice I can't see any such error 03:06 hmmmm sure the treegen code is extremely sloppy but there aren't any apparent bugs 03:06 paramat hasn't happened here 03:08 Zeno` I know what the bug is (most likely), I just can't reliably trigger it :/ 03:12 paramat new mgv7 caves is almost ready for initial review, PR incoming 03:13 hmmmm WTF is load_if_inexistent anyway 03:14 hmmmm i don't remember writing that 03:14 hmmmm i think the bug is more likely to be in the mapgen 03:14 Zeno` it's confusing I know that much 03:15 Zeno` does CONTENT_AIR ever use param1 and param2? 03:15 acerspyro VanessaE: you know we found glitchiness in your vanilla server earlier?¿ 03:15 Zeno` I'm currently testing this: https://github.com/Zeno-/minetest/commit/689708693bbb591dfba6502313c512dd67a4985e.patch 03:15 Zeno` I'd rather have memset(new_data, 0, new_size * sizeof(*new_data)); though 03:16 acerspyro oh 03:16 acerspyro That's what you guys are working on... 03:16 * acerspyro retracts himself 03:16 paramat air must use param1 for light? 03:16 hmmmm that's not fixing the bug though, zeno 03:16 Zeno` but why is VOXELFLAG_NO_DATA apparently ignored? I.e. I thought if the flage was VOXELFLAG_NO_DATA then it shouldn't even matter if the "node" is random (unitialised) data 03:17 hmmmm voxelmanipulator flags are kind of stupid in that they're never used 03:17 hmmmm it sounds like a great idea but in the actual usage of voxelmanipulator you simply find yourself not needing it... ever 03:19 hmmmm okay, so apparently I did write load_if_inexistent 03:19 acerspyro Are those garbage blocks produced possible to replicate as an item? 03:19 acerspyro Say, weirdly shaped stuff and smaller blocks 03:21 hmmmm https://github.com/minetest/minetest/commit/2c0b51795e6fa6747d881f1871c89830abb6e6e8 03:21 paramat do post any new screenshots of these crashy mapblocks, they're amusing 03:22 hmmmm waay back in 0.4.8 i made that 03:22 hmmmm yeah ewh 03:22 hmmmm i remember that bug 03:24 hmmmm so I don't think celeron's original voxelmanipulator inexisting block handling has any place at all within modern day minetest - it caused nothing but problems :( 03:25 acerspyro I had one a while ago 03:25 acerspyro If someone could check the #minetest logs 03:31 Zeno` well since they're never used that's how I introduced the bug. I should have made sure they were used :/ 03:31 Zeno` Didn't know I'd need to check that, heh 03:31 hmmmm :/ 03:31 hmmmm https://github.com/minetest/minetest/commit/714a4f85070770e81a1b0280d2c300dade9d3f7d 03:32 Zeno` except it might not be the bug 03:33 paramat hmmmm, #2123 is ready for initial review, the new tunnels look really good 03:33 ShadowBot https://github.com/minetest/minetest/issues/2123 -- Mgv7: New cave system with fewer original large caves and new 3D noise tunnels by paramat 03:33 Zeno` maybe that's it then? Because I removed my fix an added a breakpoint for when that inexistent (is that a word?) is used and I can't see any corrupt blocks (been trying for 30 minutes) 03:34 acerspyro nonexistent 03:35 paramat would it seriously break worlds to reduce lava range to 4? seems more appropriate for a thick liquid and reduces liquid queue overload 03:35 hmmmm Zeno`, I guess it makes sense. can you explain how data gets initialized there? 03:36 Zeno` initialised where? 03:36 hmmmm in initialEmerge 03:36 hmmmm if the block is nonexistant 03:37 hmmmm blitBackAll disregards all voxelflags 03:37 Zeno` yeah... 03:37 hmmmm if you do an initialEmerge and load_if_inexistent is set to false I don't see how it *wouldn't* write back corrupt data 03:37 hmmmm ..but I personally haven't seen any of these corrupt blocks myself 03:38 Zeno` I meant maybe what you changed is the real issue and I am chasing a red herring 03:38 hmmmm what you changed in your last gist is not solving the problem 03:38 Zeno` apparently not 03:39 hmmmm that just partially undoes some optimization to account for a use of uninitialized data in voxelmanipulator 03:39 Zeno` yep... Zeno`> except it might not be the bug 03:39 Zeno` I should have said "the actual bug" 03:40 Zeno` I just wish I could reproduce (I have once) 03:40 Zeno` but not when I was actually wanting it to happen >.< 03:41 Zeno` looking at the code I think your fix is the more likely the /actual/ error 03:42 hmmmm TODO: explain to me how there aren't way more corrupt mapblocks than what we're seeing now? 03:44 paramat back in 1hr.. 03:45 Zeno` hopefully paramat is off making a test case that can reproduce the bug 03:47 Zeno` hang on... 03:47 Zeno` what if a schematic is loaded into a inexistent area? 03:47 * Zeno` looks at that area of code 03:48 hmmmm nothing 03:49 hmmmm i fixed the only call to initialEmerge with that optiona boolean set 03:50 Zeno` so is that boolean actually used anywhere else? 03:50 Zeno` maybe it can be deleted hehe 03:51 hmmmm for historical reasons, i wouldn't 03:52 Zeno` maybe. It's only used in 5 places and they (since your fix) all default to true now 03:53 Zeno` 6 places 03:57 Zeno` In any case, you fix *looks* like it should work 03:57 Zeno` s/you/your 05:56 hmmmm well 05:56 hmmmm i guess we'll come back to this issue in a week or so and see if there are any more reports of the uninitialized mapblock problem 05:57 hmmmm if people think it's as interesting as they say it is, why don't they just make a mod for it... generating blobs of random nodes 06:05 paramat i think Vanessa's mods have so many nodes they colourfully vomit some up 06:06 hmmmm yeah i've always wanted to add a "colorlike" drawtype that stores an index in the available param2 bits to an RGB tint color in content features 06:07 VanessaE I've pulled to HEAD again for the server, so the attempt to fix those corrupt mapblocks will be in place after the restart (~20 mins) 06:07 VanessaE servers* 06:07 hmmmm so for example, stairs have 8 bits available to it but it has facedir which takes up 5, leaving 3 for colorlike. so you can reduce the 100 nodes of the same exact thing to more like 13 06:08 hmmmm this is super simple it's just that I never did it for some reason 06:08 VanessaE hmmmm: because there are a LOT more than 8 colors, even the base game has 15 or something. 06:08 hmmmm what? 06:09 VanessaE 3 bits for a color spec? 06:09 hmmmm if there's 3 bits available, that's 3 bits more than there used to be 06:09 VanessaE that's 8 colors, unless I'm counting bits wrong... 06:09 VanessaE you couldn't even represent the Wool mod with that 06:09 hmmmm you can reduce 100 different nodes to about 13, that's a pretty huge win in my opinion. 06:09 hmmmm i have literally no clue how you could be bitching about this. 06:10 VanessaE who's bitching? 06:10 hmmmm you... 06:10 VanessaE O_o 06:10 * VanessaE looks over at paramat 06:10 hmmmm you're complaining that this is a horrible stupid idea 06:10 hmmmm because i can only fit 8 possible values in 3 bits 06:10 * paramat stays out of it 06:10 VanessaE I'm saying it's the wrong way to handle it entirely. 06:11 hmmmm what's the right way, pray tell, vanessae? 06:11 VanessaE the only right way is one which I already know you'll reject anyway 06:11 hmmmm erm.. 06:11 VanessaE add another byte. call it param3 and make it multipurpose if you want 06:12 hmmmm having small mapnodes is a huge win over other minecraft-like games 06:13 hmmmm if it weren't for you and sapier crying about mobs, i could have 4 more bits from param1 06:13 VanessaE wat 06:14 hmmmm doubling the size of a mapnode == no 06:15 hmmmm you saw the massif output graphs... the vast majority of the minetest memory usage is MapBlocks 06:15 VanessaE look, the only way those 8 bits are gonna work out in practice is if they point to a palette that's stored in the node's metadata 06:15 VanessaE er 3 bits 06:15 hmmmm you'd double the memory consumption for one 06:15 hmmmm probably double the average map size 06:15 hmmmm double the amount of cache misses 06:15 hmmmm double the amount of time it'd take to blit mapblocks around 06:16 VanessaE 4 bytes to 5 somehow equates to double the map size? 06:16 hmmmm yup... 06:16 VanessaE did I miss a lesson in math? 06:16 hmmmm you did 06:16 VanessaE when I last checked, that was a maximum of 25% more, not 100%... 06:16 VanessaE but that's immaterial now 06:16 hmmmm ugh 06:17 hmmmm you, taoki, and mauvebic seem to keep forgetting that you can't add just *one* more byte 06:17 VanessaE like I already said before I even suggested it, I expected a rejection 06:17 hmmmm i explained this about 10 times 06:19 hmmmm the very VERY least amount of size it would be is 6 bytes in order for things to be aligned in memory and not crash. but the compiler would most likely round it out to 8 bytes for speed and memory access convenience 06:19 hmmmm so it's like 06:19 VanessaE wait wait wait 06:19 VanessaE you're speaking strictly about in-memory? 06:19 hmmmm have your 5 byte data structure get padded out to 8 and waste 3 bytes? 06:19 hmmmm of course i'm talking about in-memory 06:19 VanessaE I was speaking about the on-disk size. last I knew, databases didn't align themselves to words/longs/whatever 06:20 hmmmm on disk size is IRRELEVANT 06:20 VanessaE um 06:20 VanessaE it's quite relevant 06:20 hmmmm those parameters will most likely be 0 and compress to almost nothing 06:20 hmmmm what you would notice is the increased memory usage and cpu time 06:22 VanessaE so from 8 MB to 16. I agree that's not exactly small change, but it's hardly the gigs and gigs you're making it sound like... 06:23 VanessaE wait, strike that, I did the math wrong. 06:23 VanessaE WAY wrong :P 06:24 hmmmm so why doesn't freeminer lead the way and make it 8 bytes 06:24 hmmmm or rather, sorry, 5 bytes :P 06:24 hmmmm see how it works out for proller 06:24 VanessaE at any rate, as I said twice now I expected you'd reject the idea. 06:26 VanessaE so now you're left with those three bits. the bottom line here is that if you can't even represent the wool mod without splitting it across two nodes, it ain't gonna work, because you're going to create a massive headache for everyone. 06:26 hmmmm bullshit... 06:27 VanessaE ok, you tell me how you'd split the defs in a mod with say 25 distinct textures, in 12 colors each texture. 06:27 hmmmm nodename = "wool:wool_" .. (color_index / 8); hue = color_index % 8; 06:27 paramat use 8 tints to tint a few basic wool colours? 06:29 hmmmm hell - i'll even modify wool to take advantage of it 06:29 hmmmm i'll show you how to do it the right way 06:30 VanessaE ehm..that would not work for the wool mod. 06:30 paramat i can't resist agreeing with hmmmmm, mapnode should certainly not get bigger 06:31 VanessaE there are 15 specific colors in that mod, all the same texture, and they're all named 06:31 VanessaE e.g. wool:pink and so on 06:32 hmmmm obviously there needs to be something added for that 06:32 hmmmm maybe wool:wool^15 could be aliased to a wool node with param2 == 15 06:32 hmmmm or something like that 06:32 hmmmm and then wool:pink can get aliased to wool:wool^15 06:33 TriBlade9 Phew, lots to read 06:33 VanessaE and if you try doing this on a complex mod that supports the unified dyes palette, it's gonna fall down bad. think 12-15 textures per color times 89 colors, and the palette follows a clearly-defined color spec. 06:36 TriBlade9 Oh, is this a fight? 06:36 VanessaE nope. 06:36 VanessaE just hmmmm and I misreading each other, as usual. 06:36 TriBlade9 Oh oke :3 06:38 TriBlade9 Question. Is there a reason that sounds are unable to by polyphonic? 06:38 TriBlade9 *be 06:38 TriBlade9 I just turned my sound on awhile back, and was surprised that the node-place sound had to finish before playing again. Its a bit disorienting. 06:39 hmmmm sound is fucked 06:39 hmmmm i guess it was a good first start but it needs serious work 06:39 hmmmm try setting a couple trees on fire; openal will run out of sound buffers 06:39 TriBlade9 Got it. There's a lot of things that need refining I guess. 06:40 TriBlade9 *There are 06:40 hmmmm in order for minetest to become acceptable it needs full time developers on it but everybody here has an actual life 06:42 TriBlade9 Well, it at least needs a solid direction and agreements on what needs to be improved 06:43 TriBlade9 Someone needs to take minetest and start using it to create a playable, publicly acceptable game, and see what needs to change in the engine from there. 06:43 TriBlade9 Perhaps Voxus will help with that. 06:47 hmmmm uhh.... we know EXACTLY what needs fixing 06:47 hmmmm what we need is manpower 06:48 TriBlade9 And maybe a checklist? :P 06:48 * TriBlade9 loves checklists 06:49 Brains I did ask earlier if there was a wiki page about scut work for newbies... 06:49 hmmmm there's not a formal list of tasks... but there's a lot of performance optimization and enhancements needed on the graphics side 06:50 hmmmm sound needs a rework to be less cpu intensive (yes, it's cpu intensive somehow..), combine similar sounds, have multiple sounds play at once, etc. 06:51 * Brains was just passing by... Sleep is necessary. 06:51 TriBlade9 Sleep is an illusion. 07:04 Zeno` just reading back through the logs... 07:04 Zeno` what's the problem with adding an additional 8-bit param? 07:06 TriBlade9 Map size apparently. 07:15 Zeno` err ok 07:22 hmmmm .....no.... read up 07:22 hmmmm reading isn't hard 07:23 Zeno` speak for thyself! 07:23 Zeno` I find reading hard 07:24 Zeno` But what I read... 07:24 Zeno` Wouldn't the extra byte be more of a protocol concern than an optimisation- or memory-related concern? 07:25 Zeno` an additional octet seems fairly "low impact" to me, but perhaps I'm missing something 07:28 Zeno` brb 07:43 * VanessaE wanders off 07:44 TriBlade9 Bai Vanessa 07:44 VanessaE night :) 08:02 Zeno` TriBlade9, what's going on with gamma? 08:03 Zeno` TriBlade9, I did have options for adjusting ambient occlusion strength initially. Is that what you're asking? 08:03 Zeno` Or are you asking why I didn't use irrlicht gamma? 08:03 Zeno` I didn't use irrlicht gamma mainly because I wanted to match to the original light table 08:04 Zeno` (for the default setting) 08:04 Zeno` But... this doesn't affect your PR does it? 08:04 TriBlade9 Zeno`, I was just wondering what the precalculated values were for. 08:04 TriBlade9 And if it would cause any trouble to change them 08:04 TriBlade9 Well, I replaced the values. 08:04 TriBlade9 With the configurable strength, which is only slightly different. 08:04 Zeno` you mean the static const float light_amount[4] ... values? 08:05 TriBlade9 Yes, indeed. 08:05 Zeno` I see no need for them to remain as they are 08:05 Zeno` you're pre-calculating them now anyway... they're just not hardcoded 08:05 TriBlade9 Alright. They just seemed to be pretty "magic number"'ish, so I was worried about changing them 08:06 Zeno` I would *guess* they were done in some strange optimisation attempt, but I don't know. 08:06 Zeno` I can't see a problem with the PR 08:06 TriBlade9 I'm also unsure of the relationship/difference between the values. 08:06 TriBlade9 See as I've added -0.1f and -0.3f to try and mimic the magic number differences. Seems to work pretty well 08:07 Zeno` It's only built once 08:07 Zeno` (referring to SN's comment and your question) 08:08 Zeno` I dunno why the magic numbers were used 08:08 TriBlade9 Alright. 08:08 Zeno` they seem to be a pre-calculated gamma of 2.2 but I don't have a calculator handy 08:08 TriBlade9 I suppose I'll play around with things, but for now I think it can't get much better. 08:09 Zeno` Yeah do so. I also played around with this but kinda stopped because 0.4.11 release was too close 08:09 Zeno` "No sense wasting CPU time even on init," I don't agree with that though :P 08:09 TriBlade9 I find the idea of different games having different ao strengths, it would give a sort of subtle difference for each game. <3 08:09 Zeno` "wasting" CPU time on init is fine (IMO) 08:09 TriBlade9 *the idea intriguing 08:10 Zeno` especially since it's a readability vs. magic numbers issue 08:10 Zeno` it's not that expensive on an FPU anyway 08:11 TriBlade9 :/ 08:11 TriBlade9 C++ is so tricky to get everything just right. 08:11 TriBlade9 Sometimes I wish minetest was written in Java, then I could actually do some important stuff. 08:13 hintss ^ 08:14 Zeno` well, 9 times out of 10 (or more) if you write it the obvious way it's the best way anyhow 08:15 Zeno` modern CPUs are, in a tangential way, kind of optimised for C (and therefore C++ because it has a normative reference of C) so *shrug* 08:15 TriBlade9 Yeah. I suppose this way isn't that un-obvious either. 08:15 Zeno` Well it kind of is... but it's less "magical" than before heh 08:17 TriBlade9 Wat? 08:17 TriBlade9 You silly boi. 08:18 TriBlade9 Okay, it would appear those "magic numbers" help keep the corners from being too obvious 08:18 * Zeno` cries 08:18 TriBlade9 Doing a screenshot comparison 08:19 Zeno` There is a paper on this somewhere 08:19 Zeno` Using the exact algorithm that's implemented 08:19 Zeno` I don't have it bookmarked on this computer though :( 08:20 Zeno` the old magic numbers or the new magic numbers? :p 08:20 TriBlade9 The difference between both 08:21 Zeno` ahh ok, I see 08:21 TriBlade9 Screenies coming in a sec 08:22 TriBlade9 I think 0.2f/0.4f is best, instead of 0.1f/0.3f 08:22 TriBlade9 Makes the corners a tad darker 08:23 Zeno` just add a comment... even if it's a subjective comment like "this looks best" lol 08:23 TriBlade9 Will do 08:23 TriBlade9 I'd post a screenie, but the internet is almost dead 08:24 TriBlade9 Okay, all settings with no subtraction: https://lut.im/HbYL4rsZ/EfMST1zK 08:25 TriBlade9 0.3/0.1: https://lut.im/zwOadrkp/0pqtWCRj 08:25 TriBlade9 0.4/0.2: https://lut.im/jIMEJEKB/ri2Gi0Y4 08:31 Zeno` hmm I see 08:32 Zeno` the third one seems too strong to me 08:33 Zeno` but the first one seems wrong as well... 08:33 * Zeno` sits on the fence :D 08:35 TriBlade9 So, perhaps the second one? 08:40 kilbith yes, 2nd one is cool 08:44 TriBlade9 Alright then, no changes need to be made to the PR 09:16 * Zeno` bangs his gravel on the bench. Merge! 11:40 jin_xi clearly something is wrong: http://i.imgur.com/Vs714T5.png 11:40 jin_xi with luavoxelmanipulator it seems. 11:41 jin_xi also register_decoration always returns nil now? it does register stuff tho. 12:10 Zeno` gregorycu, why is an extent empty if the two subtractions are -1,-1,-1 ? 12:11 gregorycu Because the extent includes the area it bounds 12:11 gregorycu Um... 12:11 gregorycu The extent is inclusive of the border 12:12 Zeno` is it empty if the result is 0,0,0? 12:12 gregorycu No 12:12 Zeno` ok, can you add a comment? 12:12 gregorycu Ok 12:12 gregorycu I deleted the branch, but I will redo, lol 12:13 gregorycu Does rain stop fires? 12:15 gregorycu (So, in other words, if an area is defined by two points (1,2,3) and (1,2,3) the block at (1,2,3) is included, so the area is 1 * 1 * 1) 12:16 Zeno` I'm not really debating you, I just want it as a comment :) 12:16 gregorycu Also, I have a feeling I've found a good optimisation for your liquid troubles 12:17 Zeno` what if the subtraction is 0,0,0? what does that mean? 12:18 gregorycu You mean the result? 12:18 Zeno` yeah 12:18 Zeno` what does return MaxEdge - MinEdge == v3s16(0, 0, 0); mean? (if anything) 12:19 gregorycu It means the area is composed only by a single node, that being MaxEdge 12:19 gregorycu (I obviously didn't decided to do it like this, I just transposed the maths to make it simplier) 12:20 Zeno` I know :( 12:20 Zeno` Anyway, once a comment is added I'll merge 12:20 gregorycu It's retarded, and I don't like it 12:20 gregorycu I'll have to redo it, give me a sec 12:20 Zeno` Say it's retarded in the comment then :D 12:21 Zeno` Anyway, without a comment it makes no sense 12:21 Zeno` (although I know it's correct) 12:33 Zeno` your branch is probably not permanently deleted on github :P 12:34 gregorycu Different PR #2125 12:34 ShadowBot https://github.com/minetest/minetest/issues/2125 -- Change UniqueQueue to use a queue and a set. by gregorycu 12:34 gregorycu No, but I should be able to just take the diff and apply it 12:36 Zeno` merging #2120 soon 12:36 ShadowBot https://github.com/minetest/minetest/issues/2120 -- Fix Android crash on keyboard connection change. by KodexKy 12:54 gregorycu must_reflow concerns me 13:05 gregorycu Zeno`: #2126 13:05 ShadowBot https://github.com/minetest/minetest/issues/2126 -- Add VoxelArea::hasEmptyExtent by gregorycu 13:10 Zeno` Brand new PR? 13:10 gregorycu Yeah, the other branch is gone 13:11 gregorycu Wouldn't let me modify 13:11 gregorycu You can diff the diffs lol 13:11 Zeno` can't be gone because it's still there on github. I'll ignore that this time :p 13:11 Zeno` is the old PR closed? 13:11 gregorycu There is "gone" and "gone" 13:12 gregorycu Is a PR even a git thing? 13:12 Zeno` The comment needs updating :) 13:13 gregorycu ? 13:13 Zeno` yeah, you could have restored the old branch (I believe). It doesn't matter now 13:13 Zeno` and empty extent 13:13 Zeno` s/and/an 13:13 gregorycu Oh mother fucker 13:13 gregorycu I fixed that 13:14 gregorycu No, I have a pending change 13:14 gregorycu ... 13:14 gregorycu I swear I fixed that before commit 13:14 Zeno` heh 13:15 gregorycu Fuck git 13:16 Zeno` just update it :P 13:16 gregorycu frdoui fghdsfigbdsfgjdsfg 13:16 gregorycu What the fuck 13:17 gregorycu No 13:17 gregorycu Is that better? 13:18 Zeno` git squash -i HEAD~2 13:20 Zeno` I'll cherry pick it I guess 13:21 gregorycu 'squash' is not a git command 13:21 Zeno` eerr 13:21 Zeno` git rebase -i HEAD~2 13:24 gregorycu What's with the line endings being fucked up? 13:24 gregorycu I assume that is what it is 13:24 Zeno` nah 13:25 Zeno` anyway, I've pushed :P 13:25 gregorycu Sweet, I can go back to coding 13:25 gregorycu You know, the bit that is fun 13:27 Zeno` git is pretty good. Just takes a while 13:29 gregorycu Didn't know you could get coolaid in Australia 13:29 gregorycu Or did you drink it all 13:32 gregorycu ;) 13:44 TriBlade9 Gawddamnit, I *just* pulled master before you committed that -_- 13:45 gregorycu If it's mine, it's worth refetching 13:45 TriBlade9 :3 13:45 TriBlade9 Wait, was yours the VoxelArea one? 13:46 gregorycu Yes 13:46 TriBlade9 I mainly tweak visuals, so there's probably not much of a difference between them 13:46 gregorycu I'm only kidding, it's a slight perf improvement 13:46 TriBlade9 :P 13:46 gregorycu Bit enough to be noticable 13:46 gregorycu big 13:47 TriBlade9 Really? It didn't seem like that big a deal, at least when I glanced over it. 13:47 TriBlade9 I don't know much about C++ though, much less about optimizing it. 13:47 Zeno` merging 2102 13:47 gregorycu The devil is in the details 13:48 gregorycu The smallest things can give you perf improvements if it happens in important areas 13:48 TriBlade9 Interesting 13:48 TriBlade9 But unsurprising 13:48 TriBlade9 Zeno`, Today is performance fix day eh? :] 13:49 Zeno` I don't really believe #2102 fixes much in the way of performance, but I cannot be bothered changing the commit message :) 13:49 ShadowBot https://github.com/minetest/minetest/issues/2102 -- Performance fixes. by onkrot 13:49 Zeno` merged 13:50 TriBlade9 Damn. Pull time again -_- 13:51 Zeno` ok, no more merges from me tonight :) 13:52 TriBlade9 Oke :) 13:52 Zeno` The other PRs I haven't tested or cannot easily understand :) 13:52 TriBlade9 Wait, it's night time for you too? 13:52 TriBlade9 Which timezone are you in? 13:52 Zeno` #2116 is interesting though 13:52 ShadowBot https://github.com/minetest/minetest/issues/2116 -- Optimize vertex calculations (maths), perf improvement by nerzhul 13:52 Zeno` gmt+10 13:52 gregorycu I am +11 13:53 gregorycu Zeno is further East than me but has a less time offset 13:53 gregorycu Figure that shit out 13:53 * Zeno` looks at the sun 13:53 Zeno` makes sense to me! 13:53 Zeno` You're on DST? 13:54 TriBlade9 Trying to convert CST to GMT right now -_- 13:54 gregorycu Yes 13:54 TriBlade9 Ah, GMT+8 here 13:54 gregorycu Where is here? 13:54 gregorycu Perth? 13:54 Zeno` DST is a world war thing where coal was scarce. The idea was that people would go to bed at the same time and burn less coal. It didn't really work IMO 13:54 gregorycu Sg? 13:55 TriBlade9 China is on a single timezone through the whole country >_> 13:55 Zeno` I.e. the savings were that they would use less energy 13:55 TriBlade9 It centers on Beijing, which makes the east quite weird. 13:55 TriBlade9 *west 13:56 Zeno` gmt+10 is the same as utc+10 isn't it? 13:56 gregorycu Apart from a few seconds :) 13:57 Zeno` or maybe not. I dunno. It's kind of meaningless anyway hehe 13:58 gregorycu Anyway, I found a per improvement 13:58 gregorycu perf 13:58 gregorycu For your water Zeno` 13:58 Zeno` *my* water? 13:58 gregorycu Yep, your water 13:58 * TriBlade9 looks in Zeno`'s cup. Yep, it's quite lazy. 13:59 gregorycu Map::transformLiquids 13:59 Zeno` nope, I want nothing to do with that water/liquid 13:59 gregorycu You love it 13:59 TriBlade9 Put a ring on it perhaps? 13:59 Zeno` I didn't write it. I nothing. I see nothing. In fact, I did not even get up this mornink! 13:59 gregorycu Anyway, I'm gunna try and get some hard numbers now 14:00 Zeno` https://www.youtube.com/watch?v=UgcxGFmYyPs 14:01 Zeno` I must report this! 16:51 jin_xi hmmmm: lua voxel manipulators in latest git still broken? http://i.imgur.com/a4abRyI.png 16:52 hmmmm what do you mean by "still broken" 16:52 hmmmm they were not broken in the first place... it's not always a bug, sometimes it's the modder's fault 16:53 jin_xi i read commit log about some fix therefore still. maybe i misread. its a mod that worked well and now places blocks of random nodes 16:53 hmmmm after that specific commit? 16:53 jin_xi no also before 16:53 hmmmm sorry, I call bullshit. there's no logical way for that to happen based on what i changed. 16:54 hmmmm well, maybe the mod is wrong. why not review the mod code first for errors. 16:54 jin_xi as i said its a new phenomenon for not so new code. another thing is that register_decoration seems to return nil as of recently 16:55 hmmmm maybe shadowninja's register_ wrappers don't pass along the return value 16:56 hmmmm what is https://github.com/minetest/minetest/commit/0fd1ee03808ee8121e7b65e5c7c9ba19599d185e ?? this is a dumb commit name 16:57 hmmmm "Use the std::wstring variant of wgettext in most cases to avoid c string conversion" or something 16:57 hmmmm welp 16:58 hmmmm can't say that this wasn't my fault, he did ask me if it looked good before pushing to upstream 17:01 hmmmm why is register_biome separate from register_ore and register_decoration ... 17:01 hmmmm also this code is broken for clear_registered_biomes etc. 17:36 hmmmm https://github.com/minetest/minetest/commit/a4a6cc8eb1197648bce9bec1f0a1aa6aa4b787b7 17:56 celeron55 "I was wondering if u guys could add horses to the game and was wondering if u could make a ccomplete open source version of minetest one that is easier to edit the code. I'm wanting to make my own version but can't figure out how to." 17:57 hmmmm hah 17:57 celeron55 holy shit these people are insane 17:57 celeron55 (people who email me directly for whatever random reasons) 17:58 hmmmm i used to have a policy of not ignoring people, but how do you even respond to that without lowering your intelligence 18:00 celeron55 i also got some spam from some advertisement company to the domain contact address 18:01 celeron55 i don't even get what they are trying to sell but... oh well 18:08 rubenwardy So much forum spam :( 18:15 kilbith "We need more moderators" (C) 18:15 kilbith yes, we know 18:17 rubenwardy Better captcha would be better than more moderators 18:18 kilbith you're constantly requesting for more mods 18:19 Calinou captcha annoys everyone, it is not a solution 18:19 Calinou report the spam and we deal with it. that's all there is to it 18:24 Krock more spammers = more moderators? 18:31 rubenwardy Captcha: 1. It only needs to be on new users. 2. Moderators are very slow. Different time zones. They're only human. Today there has been 6 posts left on for at least 8 hours. 3. I meant to imply other prevention under 'captcha' such as email checking. 4. I came in on c55 talking about spam, and assumed you were talking about forum spam. ("also") 18:34 Calinou is e-mail checking done currently? 18:35 rubenwardy Don't know. stopforumspam.com is a good resource, there'll be a plugin for phpbb. 18:35 rubenwardy Although don't ban IPs - they change 18:37 Calinou banning spambots is mostly useless 18:37 Calinou we do it when we have time 18:37 Calinou (which is, often) 18:37 Calinou usually spambots only make a few posts and leave, because they know they get banned 19:37 sapier does anyone know nerzhul ? 19:58 Krock omg that pull looks.. interesting. two commits for changing the same thing just in an other way.. they should be squashed 19:59 sapier yes and coding style is a mess too but the direction is correct 19:59 Krock exactly 20:00 sapier I'm not exactly sure but it even seems to be compatible to old clients/servers 20:01 Krock it requires a good testing be4 anything is merged. I don't have an overview in those codes 20:02 Krock at least the vertex calculation looks ok. 20:22 hmmmm hum 20:23 hmmmm what's the situation with driver types and opengles on irrlicht? i noticed that the driver config strings are only present in the array being checked against when support is compiled in 20:24 hmmmm does this mean that ogles1 and ogles2 have the same enumeration value when being passed along to irrlicht? 20:24 hmmmm and what if both of them are supported on the same device? the enumeration ids change then... 20:24 sapier I don't think so 20:25 sapier I meant same enumeration value 20:25 hmmmm yeah but what if _IRR_COMPILE_WITH_OGLES1_ is set but not _IRR_COMPILE_WITH_OGLES2_ 20:25 hmmmm then ogles1 will be 5 20:26 sapier can you show me the code you're talking about? 20:26 hmmmm but if the opposite is true, ogles1 is compiled in but not ogles2, ogles2 is the same enum value of 5 20:26 hmmmm i don't know anything about android so i don't know if this is correct, intentional behavior or not 20:27 sapier let me check but for what I remember on android the enum always contains all driver types 20:28 sapier yes the enum doesn't depend on compile with flags 20:28 hmmmm i dunno about you, but my distribution of irrlicht doesn't define opengl es at all 20:28 hmmmm it goes right from EDT_OPENGL to EDT_COUNT 20:28 sapier true that's a change in current irrlicht android branch 20:28 hmmmm this is why i'm confused... the documentation also reports the same 20:28 hmmmm the online documentation 20:28 sapier if they don't merge it that way we have to adjust it 20:29 sapier irrlicht android branch isn't yet merged to main irrlicht code 20:30 hmmmm nothing happens with irrlicht 20:30 hmmmm it's such an incredibly inactive project 20:30 hmmmm probably not as bad as xfce though... 20:30 sapier true but right now we don't have an issue there, do we? 20:31 hmmmm well, no, not if _IRR_COMPILE_WITH_OGLES2_ implies _IRR_COMPILE_WITH_OGLES1_ 20:31 hmmmm i don't know if that's true or not though. 20:32 sapier right now we always compile both but you're right that might change. Do we check for those flags on driver detection? 20:32 hmmmm yes 20:33 sapier we shouldn't 20:33 sapier but wait, right now we need to check for it to decide about the enum beeing present at all 20:34 sapier we shouldn't have to check for it but as gles is inofficial that's a workaround 20:34 hmmmm shouldn't irrlicht device creation simply fail if it's passed an enum value unsupported or out of bounds 20:34 hmmmm but we don't even need to bother because the index we scan for the driver name is bounded by EDT_COUNT 20:35 sapier where's the code you could be right about this 20:36 hmmmm it's toward the bottom of main.cpp... i'm changing it anyway, not really a big issue 20:37 sapier ah ... well just remove the ifdefs but add a comment about those id's beeing inofficial 20:38 sapier wait you have to fix the for loop too 20:38 hmmmm i wanted to split it up into several functions and put it into porting.cpp with the rest of the irrlicht-related things 20:39 sapier well I guess you don't need my advice on doing so ;-) 20:39 hmmmm so you have a std::vector getSupportedDrivers() and then g_driver_name[] = { blah blah } 20:39 hmmmm this way it's shared between main and l_mainmenu 20:40 hmmmm and the options config 20:40 sapier I'd suggest adding some poritng::getDriveName(irr::core::E_DRIVER_TYPE) function 20:41 sapier this way you wouldn't have to cast driver type to int to get the name 20:42 hmmmm that's true 20:42 hmmmm i really dislike the way C++ handles enums 20:42 sapier well it's quite consequent ;-) 20:43 sapier a enum ain't an integer 20:43 sapier I'm not even sure standard requires an enum to start at 0 if it's not manually specified 20:44 hmmmm it does actually 20:44 hmmmm ... in fact i think we went through this already... 20:44 hmmmm or nevermind that was kahrl 20:44 sapier well it doesn't matter 22:26 ShadowNinja updatepo.sh tweaks, fixes #2053: http://sprunge.us/JGRQ?diff 22:27 ShadowBot https://github.com/minetest/minetest/issues/2053 -- German Ingame-Translation is broken in 0.4.11 22:27 ShadowNinja Also uses long options since they're 10x more readable. 22:28 ShadowNinja I could split it. 22:57 VanessaE hmmmm: your attempt to fix the corrupt blocks failed. Unless the reporter is wrong, this is all in new terrain today: http://imgur.com/0r8whs9 23:01 TriBlade9 Huh, there's no way in Lua to set the width of an extruded weildmesh is there? 23:02 TriBlade9 I'll have to take a look into that, could be quite useful for items like bricks or lumps. 23:08 TriBlade9 Nvm, there already is apparently. 23:20 VanessaE [01-13 18:20] *sigh* i just watched another one appear before my eyes >.< 23:20 VanessaE come on guys, this shit is destroying my maps 23:23 hmmmm oh, let me just magically come up with the solution 23:23 hmmmm we're just not trying hard enough 23:23 VanessaE um.... more like you're not *testing* enough 23:24 VanessaE I've got *one* guy finding these left right and sideways 23:24 hmmmm um.. i am. 23:24 VanessaE then how come he finds them and you don't? nevermind if you can fix the problem -- I get that -- you keep saying you never even *see* them happening 23:24 hmmmm why don't you do something useful like give us the mods he was using and try instructing him how to isolate the offending mod instead of complaining about how we're not working hard enough 23:24 VanessaE Dreambuilder Game. 23:24 hmmmm that's an entire game. 23:25 hmmmm a game is composed of one or more games. 23:25 hmmmm mods. 23:25 VanessaE he's not "using" a mod. he's just exploring new terrain 23:25 VanessaE that's all. 23:25 hmmmm you're using a mod if you're exporing new terrain any single one of those mods calls register_on_generated(). 23:25 VanessaE so pretty much the only things that kick in at that point are plantlife and moretrees 23:25 VanessaE and those don't use lua-side vmanips 23:26 hmmmm so then disable moretrees and see if that fixes the problem. we'd know exactly where to look if it does. 23:26 VanessaE I believe zeno already went down that road 23:27 hmmmm and? 23:27 VanessaE he got it down to just minetest_game + plantlife + moretrees as being enough to at least intermittently cause the failure 23:27 hmmmm how about removing plantlife just to be absolutely sure 23:29 VanessaE impossible. 23:29 VanessaE plantlife does not use vmanip at all. 23:29 hmmmm again: WHO THE HELL SAID IT WAS VMANIP 23:29 VanessaE um, you did? 23:29 hmmmm i didn't 23:29 hmmmm it could be, it might not be 23:30 hmmmm why does everything have to be an argument with you 23:30 VanessaE well what else can it be? at this point the only operations being performed that write to the map are set_node() and spawn_tree() 23:30 hmmmm why don't you try doing something different in your life aside from minetest 23:30 hmmmm chill the fuck out. 23:30 VanessaE wat 23:30 hmmmm jeez. 23:30 * VanessaE wanders off 23:30 VanessaE enough of this shit 23:30 hmmmm in any case it could be either spawn_tree() or set_node() 23:30 hmmmm i need to look at both. 23:48 jin_xi hm. i get deco ids starting from -1. 23:49 hmmmm that means registration failed 23:49 jin_xi oh good