Time Nick Message 00:22 paramat RealBadAngel see new comments in 3695 00:23 RealBadAngel ok 00:26 RealBadAngel paramat, ok with space after "if", but im using brackets for every oneliner if, for or whatever 00:26 paramat ok no problem 00:27 RealBadAngel that makes "harder" to make stupid mistakes when modyfing or reading code later on 00:31 RealBadAngel paramat, fixed 00:31 paramat unit tests failed in one of the builds, lets see what re-triggering the builds does 00:34 RealBadAngel those unit tests are pretty funny 00:34 RealBadAngel i saw it starting all over several times one day in a PR 00:34 RealBadAngel once failing once passing 00:39 Hijiri actually it seems like non-fleshy damage types are working after all?? 00:39 Hijiri I'm not sure what's happening 00:42 Hijiri ok 00:42 Hijiri I think I see 00:42 Hijiri the reason I thought it wasn't working was because 3d_armor mod was healing the player before I could notice 01:12 RealBadAngel paramat, https://travis-ci.org/minetest/minetest/builds/109250373 01:12 paramat good ok 01:20 RealBadAngel after that please close issue #3702 01:20 ShadowBot https://github.com/minetest/minetest/issues/3702 -- Liquids Rendering issue 01:21 paramat ok 01:35 sofar paramat: you're making the game too easy :) 01:39 paramat could make wheat seeds only dropped by junglegrass, that can be a 5000 node walk 01:40 sofar having them drop from normal grass is fine 01:40 sofar I just don't like that with hands they drop grass nodes 01:40 sofar it would actually make swords useful in minetest_game 01:41 RealBadAngel to harvest seeds? thats indeed fine usage for swords :P 01:41 sofar no 01:41 sofar opposite 01:41 sofar without swords you can't get leaves, grass or junglegrass 01:41 sofar with swords you will never get seeds 01:41 sofar just grass or leaves 01:41 RealBadAngel still 01:42 RealBadAngel better make a machete for that ;) 01:42 sofar I was suggesting something like that 01:42 sofar but swords are already in the game, and that's fine as well 01:43 sofar the actual tool isn't pertinent 01:44 RealBadAngel but nobody saw ever farmers running around with swords to get some seeds anyway ;) 01:44 RealBadAngel or a few leaves 01:45 sofar it's not the important part of the change needed 01:48 sofar introducing a machete without the other part will never get accepted, anyway 01:51 VanessaE that said, don't most folks who DO go out roughing it use an axe or hatchet to gather leaves? 01:52 sofar why would you? it wears down the tool and hands have the same result 01:52 sofar that's exactly my problem with it :) 01:52 VanessaE not saying you should in-game 01:52 VanessaE rather that it doesn't matter if it's an axe or a sword or something else 01:53 sofar axe is fine too 01:53 VanessaE though I guess one wouldn't even begin to consider using a shovel ;) 01:53 sofar I honestly don't care, as lnog as it's a sharp tool :) 03:13 jhcole sofar: I really like the new doors api but I'm having trouble with backface_culling. It doesn't seem to be working. is this just me? 03:23 RealBadAngel huh, got those nametags ALMOST working 03:24 RealBadAngel i do have all the nametags (pos and text) already on HUD side only 03:25 RealBadAngel now the black magic :) 03:25 RealBadAngel btw, with this i will solve at least 3 issues lol 03:26 RealBadAngel nametags hidden behind water, opaque halo, and all the liquids transparency issues 03:30 sofar jhcole: update minetest 03:31 sofar RealBadAngel: can you fade nametags with distance? 03:31 RealBadAngel sofar, easily if requested 03:32 jhcole sofar: i'm built from commit 3a74b84 03:32 sofar jhcole: what exactly is the problem? 03:34 jhcole sofar: the backface is visible 03:34 sofar on which door? 03:34 jhcole sofar: for wood and steels doors 03:34 jhcole and a door i'm trying to add via the api 03:35 sofar did you make sure to restart the server? 03:35 jhcole yeah 03:35 sofar screenshot (fullscreen pls, with f5 debug enabled) 03:37 sofar well, it's still working for me. did you modify door code or anything? can I see your door registration def? 03:39 sofar updated myself to 3a74b84 and it's still fine 03:39 sofar jhcole: ? 03:40 jhcole sofar: i'm working on getting a good screenshot 03:40 jhcole sofar: just a sec 03:43 sofar you know f12 works, right? 03:44 jhcole sofar: http://imgur.com/kDZeQGu 03:45 jhcole i wanted to get the scene right 03:45 sofar so what version of minetest is running on the server? 03:45 jhcole i just built the latest commit tonight, its running in place 03:46 jhcole latest minetest and latest minetest_game 03:46 sofar and the client is dirty? local modifications to the tree? 03:47 jhcole i think its saying that because there are a few archive files inthe root dir from previous builds 03:47 sofar are you running the client from the same in place location? 03:48 jhcole oh sh*t 03:49 sofar well no, poop is most likely not the reason 03:49 jhcole no my client is out of date, running wrong machine 03:49 sofar yes, old clients will not see culled mesh nodes 03:50 jhcole cool, I wish i would have thought of that :) 03:50 sofar no worries, you're not the first today that claimed my code was broken :) 03:51 RealBadAngel http://i.imgur.com/fWUp96b.png 03:51 * RealBadAngel yawns 03:51 sofar navel name tag 03:52 RealBadAngel its in HUD now 03:54 RealBadAngel ha, it will be possible to make nice screenshots on servers without all those nametags around 03:54 jhcole sofar: I actually think its a great improvement. I'm trying to update the landrush shared_door to use the new api, but i found i need do some checking to see if the door's area is shared, so i modified the door api to accept an on_rightclick in def 03:55 sofar show me those changes, I'd like to see them 03:55 sofar (I have a few door fixes already pending) 03:55 sofar I want to make sure we keep the code cleaned up well 03:55 jhcole sofar: ok, i'll push in a few minutes and link 04:06 sofar jhcole: this for a subgame, I suppose? 04:07 jhcole sofar: i guess so, what the definition of a subgame? its for the LLM II server at 162.211.32.154 04:09 sofar well if you're patching minetest_game you either submit those changes for merging or you are maintaining a subgame :) 04:21 RealBadAngel so,http://i.imgur.com/IFoFJRO.png 04:22 RealBadAngel finished, will clean and push that after a nap 04:22 jhcole https://github.com/FozLand/landrush/commit/096b39518241e594d14e4c73063b766e47600968https://github.com/FozLand/minetest_game/commits/doors 04:23 jhcole that was supposed to be two separate links https://github.com/FozLand/landrush/commit/096b39518241e594d14e4c73063b766e47600968 and https://github.com/FozLand/minetest_game/commits/doors 05:03 paramat !tell rubenwardy see new comment in #3140 05:03 ShadowBot paramat: O.K. 06:03 paramat now merging #3695 06:03 ShadowBot https://github.com/minetest/minetest/issues/3695 -- Use vertices with tangents only when its needed. by RealBadAngel 06:12 paramat merged, and closed issue 3702 06:20 sofar nice, it feels slightly faster 06:39 paramat i've just tested with camera offset step of 31000 and made snow nodeboxes 1/16th node thick, i only noticed them blinking at roughly 1000 nodes altitude 06:56 paramat hm with 1/32th node nodeboxes they start to get bad at 200 nodes distance, so the current setting may be necessary 07:00 paramat .. depending on how common 1/32th nodeboxes are 07:06 paramat retesting, even 1/16th nodeboxes start to get glitchy at 200 nodes 07:08 paramat so setting seems good, closing issue 07:12 RealBadAngel paramat, if only one specific drawtype is wrong, that means nothing other that theres something wrong with the drawtype not the camera 07:16 RealBadAngel so, what about fading nametags with distance? 07:17 RealBadAngel shall i add it? hmmmm, kahrl, nore, sfan5 ? 07:17 VanessaE RealBadAngel: PLEASE DO. 07:17 sofar yeah man, please do 07:17 RealBadAngel any ideas on actual distance? 07:18 VanessaE fade to the same distance as the current view range 07:18 VanessaE maybe go say 100m past that. 07:18 VanessaE I shouldn't see a player's name 2000 nodes away if I can only see out to 500 or whatever 07:19 VanessaE well, maybe have a fade limit 07:19 sofar I would fade at math.max(camera distance, 100) 07:19 sofar wheps 07:19 VanessaE barely readable once they get out past the fog, but no more faded than that. 07:19 VanessaE (if that makes sense) 07:19 RealBadAngel remember that im fixing rendering distance too with another commit, so actual vRange will be a REAL limit of what you can see 07:19 sofar I would fade at math.min(camera distance, 100) 07:19 sofar the lower of 100 or camera distance 07:20 sofar don't even bother going past camera distance 07:20 sofar or more than 100 nodes 07:20 VanessaE 100 nodes might be a bit short. 07:20 RealBadAngel http://i.imgur.com/YyiUNzZ.png 07:20 VanessaE it only takes a matter of seconds to run that far 07:20 RealBadAngel thats 40 nodes distance 07:20 sofar I don't think you can see players from that far 07:21 VanessaE that's fair - if you can't discern the actual player model, you shouldn't be able to hardly see the name at that point 07:21 RealBadAngel fog here is adjusted to match the view distance 07:21 VanessaE player models themselves ought to fade too, but I don't think they do (do they?) 07:21 VanessaE (I don't remember) 07:21 RealBadAngel with my changes they do 07:21 VanessaE RealBadAngel: got a before-and-after of that 40m range? 07:21 RealBadAngel without theyre always visible 07:21 VanessaE I'm having trouble mentally comparing it 07:22 RealBadAngel ok 07:22 VanessaE of course there probably should be a config option to disable the fading behavior 07:23 VanessaE (even better: hide nametags when HUD is hidden) 07:23 RealBadAngel but first: player at edge of 40m http://i.imgur.com/2RC5wel.png 07:23 VanessaE I see it. barely :) 07:24 RealBadAngel VanessaE, theyre part of hud now, so its obvious theyre gone when you turn hud off 07:24 VanessaE how about the distance limit of a nametag be double your current view distance, or 100m, whichever is more. 07:24 RealBadAngel can try 07:25 VanessaE [02-15 02:23] I would fade at math.max(camera distance, 100) <-- camera distance * 2 07:26 sofar why show name plates for players well outside visual range? 07:26 VanessaE so that you can follow them 07:27 sofar it makes pvp way too easy 07:27 VanessaE maybe. 07:29 RealBadAngel it can be optional, so pvp servers could turn fading on 07:30 sofar well so, the easiest way is to make less configure options that can achieve more ways of configuring 07:30 RealBadAngel also we can toggle it with a key 07:30 sofar don't make it client-abusable 07:30 RealBadAngel if server wont allow it wont work 07:30 sofar allow the server to set it? 07:30 RealBadAngel same as with minimap 07:31 sofar server should be able to set the max 07:31 sofar clients should be able to turn off plates and choose lower range? 07:31 sofar max = 0 -> server forces plates off entirely 07:32 sofar max = -1 -> server allows plates on range at any level 07:32 sofar that way everything is possible (except make it camera distance dependent 07:32 sofar you could reset the client to some camera dependent value, though 07:33 RealBadAngel i think we shall leave that for discussion because it can really affect gameplay 07:34 RealBadAngel im already calculating distance to the point so applying it is easy 07:34 RealBadAngel (code is using same math as waypoints) 07:48 kahrl RealBadAngel: regarding #3705, I have figured out a possible reason where the 1.1 factor came from 07:48 ShadowBot https://github.com/minetest/minetest/issues/3705 -- Fix too agressive block culling by RealBadAngel 07:48 kahrl see https://gist.github.com/kahrl/379e39767da95ff16bc7 07:48 kahrl (octave code) 07:49 kahrl fov_x/2 is the angle between the center of the screen and the top center of the screen 07:49 kahrl (or bottom center) 07:50 kahrl similarly, fov_y/2 is the angle between the center of the screen and the left center of the screen (or right center) 07:50 kahrl but the angle between the center of the screen and a *corner* of the screen is larger than both of these 07:50 kahrl my code in that gist computes that angle 07:51 kahrl isBlockInSight uses max(fov_x/2, fov_y/2) with some minor adjustments for culling 07:51 kahrl therefore, it sometimes culls mapblocks that are in the corner of the screen 07:52 nrzkt2 RealBadAngel, kahrl i didn't understand what this PR is for 07:52 kahrl as I computed, the angle to the corner is 1.10421165985322 as large as max(fov_x/2, fov_y/2) 07:53 RealBadAngel kahrl, so it looks like my empirical 1.1 is right :) 07:53 kahrl nrzkt2: load a bunch of terrain, press r to enable full range, then fly above the terrain 07:53 RealBadAngel r is not needed even 07:53 kahrl nrzkt2: sometimes mapblocks in the corner of the screen get culled 07:53 kahrl RealBadAngel: yeah 07:53 RealBadAngel fly up 100 nodes 07:54 RealBadAngel look down 07:54 kahrl RealBadAngel: but without it the culling may be because the mapblock is out of range, so r avoids that 07:54 RealBadAngel and around while still lookin down 07:54 kahrl RealBadAngel: well, the 1.1 is right with the aspect ratio and fov_y I used in that computation 07:54 RealBadAngel ofc 07:54 kahrl RealBadAngel: like I mentioned at the end, with other inputs the factor can get close to sqrt(2) 07:54 nrzkt2 the translation for culling is very strange in fact :p 07:55 kahrl RealBadAngel: but we definitely don't want to multiply the angle by sqrt(2), that's waaay to large 07:55 kahrl too* 07:56 RealBadAngel need to go out now, will you make PR with proper math? 07:56 kahrl I might try, not sure if I'll succeed 07:57 kahrl I've yet to understand the math behind adjdist 08:01 RealBadAngel ok 08:03 RealBadAngel just a thought maybe modifying my magic number with aspect ratio will do 08:05 RealBadAngel less math, faster 08:07 RealBadAngel this is done per each mapblock in range every frame, so less math the better 08:10 kahrl I think in general much of isBlockInSight could be precomputed once per frame 08:13 kahrl maybe isBlockInSight should first do a very coarse check (that never falsely culls a mapblock, but may fail to cull some mapblocks that could be culled) 08:13 kahrl then if a mapblock passed that test, do a more precise test 08:13 kahrl perhaps something like http://www.gamedev.net/topic/512123-fast--and-correct-frustum---aabb-intersection/ 08:15 kahrl hrmm, I just noticed that my fov_x calculation wasn't entirely correct 08:16 kahrl the correct one should be: fov_x = atan(aspect_ratio * tan(fov_y / 2)) * 2 08:16 kahrl but if I do the the factor comes out as 1.14423462298424 08:16 kahrl if I do that* 08:18 Megaf ShadowNinja, #3699 =/ 08:19 ShadowBot https://github.com/minetest/minetest/issues/3699 -- Use SQLite's R*Tree to store rollback node positions by ShadowNinja 08:19 paramat about nametag distance, we have a 'player_transfer_distance' setting that makes it settable 08:19 kahrl maybe #3705 was "close enough" to 1.14423462298424 if you count in the adjdist stuff and so on, so that it worked 08:19 ShadowBot https://github.com/minetest/minetest/issues/3705 -- Fix too agressive block culling by RealBadAngel 08:37 Megaf folks, minetest's cmake can't find any jpeg here. Could NOT find JPEG (missing: JPEG_LIBRARY JPEG_INCLUDE_DIR) 08:38 Megaf but I have all jpeg libs it could ask for http://paste.debian.net/plain/388375 08:39 kahrl Megaf: do you have the include files for libjpeg? 08:41 Megaf -DJPEG_LIBRARY=/usr/lib/x86_64-linux-gnu/libopenjpeg.so.1.5.2 didnt help either 08:41 Megaf kahrl, I don't know, it always worked... well, worked months ago when I last played Minetest, I've been using only my fork, that doesn't have jpeg support at all.... 08:42 Megaf -DJPEG_LIBRARY=/usr/lib/x86_64-linux-gnu/libopenjpeg.so.1.5.2 -DJPEG_INCLUDE_DIR=/usr/lib/x86_64-linux-gnu/ 08:42 Megaf this seems to have worked... 08:43 kahrl huh 08:43 kahrl why would there be include files in /usr/lib 08:43 Megaf you are the minetest dev, you tell me :P 08:44 kahrl well I guess there aren't, but cmake might skip checking for them if you pass it all the paths and assumes you know what you're doing 08:45 kahrl and you might get a working build since minetest never actually includes jpeglib.h 08:48 Megaf kahrl, IMHO jpeg support could be disabled entirely 08:48 Megaf disabled/removed 08:48 kahrl you typically can't 08:49 kahrl if you install irrlicht through your distro's package manager it includes jpeg loading support, so it has to be linked with the jpeg lib 08:50 kahrl the only way around that would be to compile irrlicht without jpeg support (or let the distro do that, if you're gentoo or similar) 08:50 kahrl you're using* 08:51 Megaf I did in my fork, didn't notice any drawback in any server 08:52 Megaf you typically can't 08:52 Megaf I did in my fork, didn't notice any drawback in any server 08:52 kahrl well, the server doesn't load images anyway 08:52 kahrl jpeg is only required if you set BUILD_CLIENT=1 08:53 kahrl oh, you mean you built a client without it and it works on any server? 08:54 Megaf sorry, I think I'm not making myself clear. So, I forked Minetest, and removed bits of code related to JPEG, so basically, it will no longer check and use JPEG. Doesn't matter if server or client 08:54 Megaf yep, that's what I meant 08:54 kahrl yeah, almost all textures are .png 08:54 kahrl I wouldn't say that 100% of them are though 08:55 Megaf sure enough I did in a imprecise and somewhat incorrect way since I'm not a developer and I don't fully understand MT code. But still, works quite well 10:40 kilbith since 0.4.13 is 6 months old, can we start talking about a new release ? 12:40 sfan5 Megaf: you'll still need to link to jpeg if irrlicht links to it though 13:01 RealBadAngel #3712 13:01 ShadowBot https://github.com/minetest/minetest/issues/3712 -- Move object nametags to HUD by RealBadAngel 13:02 RealBadAngel est31, kahrl, nrzkt2 can you take a look? 13:06 est31 what the hell 13:06 est31 object nametags have nothing to do with HUD 13:06 est31 they dont belong there, not at all 13:07 est31 RealBadAngel, is there any actual reason to do this 13:07 RealBadAngel lots of reasons, you will be suprised :) 13:07 RealBadAngel remember water going opaque? 13:07 RealBadAngel or halo? 13:08 RealBadAngel irrlicht text scene node is bugged and is messing with materials 13:09 RealBadAngel thats a). b) is that text is not a part of world geometry. otherwise it will be shaded like everything. and will cause depth frenzy, text dont write to depth buffers 13:10 RealBadAngel even if text node wont be buggy it will have to go 13:10 est31 even if it interfered with shaders, which is a legitimate reason, it should not be managed by HUD 13:10 RealBadAngel it should, its text displayed on HUD, like everything else 13:10 est31 HUD is for stuff not related to the env 13:11 est31 stuff thats about the player itself 13:11 RealBadAngel its info shown above the player 13:11 est31 yes but other players not you 13:11 RealBadAngel not a texture, not a real object 13:11 RealBadAngel info is show to you 13:11 RealBadAngel *shown 13:12 est31 info is shown to all 13:12 est31 also, when I hide the hud the nametags should stay 13:12 est31 so -1 from me 13:12 RealBadAngel so each player can see it on own hud 13:12 RealBadAngel est31, are you aware how many issues that PR fixes? 13:12 est31 like? 13:12 RealBadAngel hold on 13:14 celeron55 i don't think it's actually important whether the dictionary definition of HUD includes those or not 13:14 celeron55 focusing on the quality of the implementation makes more sense 13:15 celeron55 it looks quite straightforward to me which means it might just as well be there; but this isn't a full review 13:17 RealBadAngel cant find the issues, those are years old 13:17 RealBadAngel mainly, it broke everything transparent in multiplayer 13:18 RealBadAngel so water becoming randomly opaque, halo etc 13:18 RealBadAngel kilbith grouped them all in one open issue, just cant locate it now 13:19 RealBadAngel celeron55, my goal atm is to clean world renderer from things that dont belong to it, texts are not a part of the world, whatever one can say 13:20 est31 texts aren't part of the HUD either 13:20 est31 next you want to draw particles via the HUD? 13:20 RealBadAngel imho the HUD is more than a proper place for it 13:20 est31 or meshgen? 13:20 RealBadAngel ofc not 13:21 RealBadAngel particles are world object 13:21 RealBadAngel can be shaded, have depth etc 13:22 RealBadAngel please do try to understand that, its not like i want to move them just for sake of moving 13:22 RealBadAngel i do have valid reasons to do so 13:23 RealBadAngel that , to my suprise, it fixes shitload of bugs meanwhile is just side effect 13:23 est31 it would be cool if you were a bit more transparent 13:23 est31 and if we found a better place to put them than HUD 13:23 RealBadAngel est31, sorry but i explained many times here what im doing and what for 13:24 RealBadAngel im fixing all the stuff that is made wrong way to get deffered renderer workin 13:24 RealBadAngel you can check channel logs 13:25 est31 you can check channel logs <---- sorry its too noisy I want to do other stuff than minetest as well 13:26 RealBadAngel i can make a roadmap of what i have to do if you wish 13:26 RealBadAngel im not forcing anything 13:26 est31 Put it on gist.github.com or somewhere you can edit it 13:26 est31 it would help :) 13:26 RealBadAngel sure, np 13:27 celeron55 i agree with moving the tags out from the world rendering even while i probably disagree with everything RBA is going to do after cleaning it up 8) 13:27 RealBadAngel celeron55, would you disagree with real lighting? 13:27 RealBadAngel it wont be any slower than what we have now 13:28 celeron55 i don't believe you can make it work reasonably for caves and non-outside places 13:28 RealBadAngel im planning to mix both systems 13:28 celeron55 but if you do make it actually work, then i might not disagree with it 13:29 celeron55 but until that happens, i'm going to be your biggest roadblock in getting it upstream 13:29 RealBadAngel we will see, theres lots to do before i can make it real 13:29 RealBadAngel celeron55, sure thing 13:29 celeron55 i'm actually curious in seeing how it works once it does 13:29 RealBadAngel im just coding things the best way i can 13:29 celeron55 because there have been many theories around here about how to go about it, but nothing has been actually done 13:30 RealBadAngel you must admit that im doing also other things on my way, ones that wasnt fixed since ages, so it goes for good, no matter the end result 13:31 RealBadAngel yup, there were many theories, but doesnt it look like i have a plan with what im doing lately? 13:31 celeron55 i don't really care 8) 13:31 RealBadAngel hehe 13:32 RealBadAngel we will see the end result, ok? 13:32 celeron55 i'm just looking at the end results 13:32 RealBadAngel thats a sure thing, but look also at what im fixing going there 13:32 est31 just it would be best to only have end results in master 13:32 est31 not like with stuff like weather 13:33 RealBadAngel atm everything im touching ends with fixing long time issues 13:33 nrzkt2 est31 you follow me :p 13:33 est31 nrzkt2, lol no 13:33 RealBadAngel texture tear, glitched wielded, transparency bugs, too agressive occlusion 13:34 nrzkt2 yes irrlicht devs should switch into github, stop using MANY \r in their code and stopping doing SVN : 13:34 nrzkt2 using* 13:34 nrzkt2 :p 13:34 RealBadAngel also camera fixes gave +10% fps for free 13:34 est31 nrzkt2, I looked up the github mirror because I've wanted to find out something 13:34 est31 nrzkt2, not related to you 13:34 nrzkt2 :p 13:34 nrzkt2 no problem ^^ 13:35 est31 nrzkt2, full agree, there are many things irrlicht could improve. 13:35 RealBadAngel est31, but back to HUD or not to HUD ;) what in your opinion is the best place for text floating around the world? 13:36 nrzkt2 but est31 irrlicht 1.9 is better than 1.8, i use irrlicht 1.9 from april 2014 and it works better than irrlicht 1.8... 13:36 RealBadAngel ive stated my pros, now your turn 13:36 nrzkt2 one of their performance improvemnet in 1.9 !! is to use function with references instead of copy... 13:37 RealBadAngel till now you just said no, because no 13:37 RealBadAngel that not a valid reason imho 13:38 est31 nrzkt2, we release packages for multiple linux distros, thats how most of the "normal" users come from. with "normal" I mean those who use minetest directly not one of its ripoffs (their count is highest). 13:38 est31 nrzkt2, and the linux distros dont have 1.9 packaged 13:38 est31 bc there is no official release 13:39 nrzkt2 est31, i know 13:39 nrzkt2 i only give your my feeling about 1.9 :p 13:43 RealBadAngel est31, so? 13:45 Fixer for people who asking what it should fix: https://cloud.githubusercontent.com/assets/16494741/13029821/5b9e831a-d29f-11e5-8953-fb81bcf8f6ce.png and https://i.imgur.com/RH4AiLP.png 13:46 Fixer also flowing water opacity fix 13:46 RealBadAngel and halo going opaque... 13:47 est31 Fixer, whats the bug in the second image? 13:47 Fixer est31, correct one is this https://i.imgur.com/vYsSnSr.png 13:47 Fixer est31, windows going from transperent to opaque white 13:47 est31 ah 13:48 est31 RealBadAngel, so you basically work around a bug in irrlicht? 13:48 Fixer est31, you can test it on Liberty Land, coordinates are in those pics 13:48 RealBadAngel est31, i told ya you will be suprised what it fixes ;) 13:49 RealBadAngel est31, basically yes. i did two things at a time. achieved my goal and fixed long standing issues at the same time 13:49 est31 RealBadAngel, congrats for finding the bug 13:49 est31 the reason for it* 13:50 RealBadAngel that was also an open issue about nametags dissapearing behind water 13:50 RealBadAngel fixed too 13:50 RealBadAngel so, good deal or not? :) 13:51 est31 RealBadAngel, if it indeed fixes the bugs, then its really good 13:51 est31 still, HUD is a bad place for it 13:51 est31 RealBadAngel, where does Hud::drawNametags get called? 13:51 RealBadAngel drawscene 13:51 est31 RealBadAngel, also, can you update the PR description on github with the issues the PR fixes? Then people know in e.g. 6 months what the PR is about 13:52 RealBadAngel sure, i just uploaded it 13:52 RealBadAngel then got involved in discussion 13:52 est31 drawscene... cant find it in the PR's diff 13:52 RealBadAngel drawscene needs also a small cleanup, theres too much repetitve code 13:53 RealBadAngel ouch 13:54 nrzkt2 RealBadAngel, agree 13:55 RealBadAngel est31, added 13:56 RealBadAngel atm draw this draw that is called here and there 13:56 RealBadAngel there should be just one call to draw all the hud stuff 13:56 RealBadAngel folks adding there different 3D modes were just copying parts of code all around 13:57 RealBadAngel est31, still you havent answered: if not HUD then what? 13:58 est31 RealBadAngel, putting it into drawscene? 13:58 RealBadAngel drawscene is a collection of calls to elsewhere 13:58 est31 idk, there should be a better place for it than HUD 13:59 RealBadAngel that wasnt any good idea just a blind shot 14:00 RealBadAngel but you are partially right. drawscene makes an order how the things are displayed 14:00 est31 RealBadAngel, for example do it like this: 14:00 RealBadAngel first the world, then all the overlays 14:00 est31 drawscene has a method, like draw_nametags 14:00 RealBadAngel wielded, hud 14:01 est31 and the method has a param which gets the list of Nametag objects 14:01 RealBadAngel following that you shall move there also waypoints 14:01 RealBadAngel infotext 14:02 RealBadAngel drawscene is not a method to draw specific things just to order when they are drawn 14:02 RealBadAngel and how 14:03 est31 its a file 14:03 est31 not a method 14:03 RealBadAngel display world, display overlays, done. or display world for left eye, right, overlays done. 14:04 RealBadAngel not to display a text god knows where 14:04 RealBadAngel im treating it as a method 14:04 RealBadAngel its primary role is to display things in correct order 14:05 RealBadAngel not to draw specific things 14:05 RealBadAngel it will command all the deffered rendering stuff, mrt and pp 14:06 RealBadAngel its simply not a right place for fancy unimportant stuff like nametags 14:07 est31 okay, I think I know now where it belongs 14:07 RealBadAngel go on 14:07 est31 camera.cpp 14:08 RealBadAngel what? 14:08 est31 its not the perfect place for it 14:08 est31 but much better than hud.cpp 14:08 est31 camera.cpp has already drawWieldedTool 14:09 est31 its the place where to put "small stuff" that doesnt deserve its own class 14:09 est31 ermm file 14:09 RealBadAngel and thats one of universe mysteries already why it has it ;) 14:09 est31 so thats where it should go 14:10 RealBadAngel because somebody put an engine in your doors, you will claim that fuel tank should also go there? 14:11 RealBadAngel sorry, but it looks like youre defending your "no" at all cost ;) 14:11 est31 RealBadAngel, sorry but it does not belong into hud.cpp 14:12 RealBadAngel thats your opinion, lets just wait for others, ok? 14:12 est31 RealBadAngel, your opinion is to have it in hud.cpp. 14:12 est31 lets wait for others 14:12 RealBadAngel yes 14:12 RealBadAngel and c55 one is also here 14:13 RealBadAngel i think im makin things consistent, see waypoint are HUD 14:14 RealBadAngel also, have you ever tried to make a screenshot of your structure on skyblock server? 14:15 est31 RealBadAngel, i dont care about that server 14:15 RealBadAngel propably not, because you would be pissed about all those nametags ruinin your nice shot ;) 14:15 RealBadAngel with them in HUD just press F2 14:15 est31 RealBadAngel, nametags have _nothing_ to do with HUD 14:15 est31 HUD means HEAD UP DISPLAY 14:15 RealBadAngel yes, EXACTLY 14:15 est31 man, I dont have any time to pointlessly fight with you 14:16 est31 do whatever you want 14:16 est31 -1 if it goes to hud.cpp 14:16 est31 waypoints dont belong there either 14:16 RealBadAngel and what nametag is? part of players body or what? 14:16 est31 but thats already old code 14:30 RealBadAngel celeron55_, could you also state your opionion in nametags PR? 14:31 RealBadAngel asking for it because it seems its gonna be hard discussion 14:40 RealBadAngel btw, https://github.com/minetest/minetest/issues/2778 14:41 RealBadAngel this can propably be closed, its about removed code 14:42 RealBadAngel https://github.com/minetest/minetest/issues/2687 also outdated 14:43 RealBadAngel https://github.com/minetest/minetest/issues/2557 same 14:44 RealBadAngel https://github.com/minetest/minetest/issues/2516 we do have nametags working, can be closed 14:46 RealBadAngel https://github.com/minetest/minetest/issues/2450 we do have tonemapping 14:49 RealBadAngel https://github.com/minetest/minetest/issues/2080 halo is here 14:50 est31 RBA #2516 is about hover text 14:50 ShadowBot https://github.com/minetest/minetest/issues/2516 -- Ability to change hover text for objects and mobs 14:50 est31 not nametags 14:50 est31 hover text is the stuff you get if you hover on the item 14:51 est31 and about #2557 I dont think its outdated 14:51 ShadowBot https://github.com/minetest/minetest/issues/2557 -- Wielding items that use minetest.inventorycube causes an error 14:51 RealBadAngel object and mobs? 14:52 RealBadAngel those are nametags 14:52 RealBadAngel check mobg 14:52 RealBadAngel mobf 14:52 RealBadAngel this mod allows you to name your pets 14:53 RealBadAngel please read the issue 14:53 RealBadAngel somebody used "hover" propably wrong way 14:54 est31 lol 14:54 est31 you dont read it 14:54 est31 quoting the issue 14:54 est31 "it always displays "GenericCAO"" 14:54 RealBadAngel " it always displays "GenericCAO" and armor/health status" 14:54 est31 thats the hovertext 14:54 RealBadAngel yes exactly 14:54 est31 _not_ the nametag 14:54 est31 the nametag is displayed always 14:54 RealBadAngel but if it has nametag you will see the nametag 14:55 est31 the hovertext is displayed only if you hover the thing 14:55 RealBadAngel not the debug info 14:55 est31 ermm what? 14:55 RealBadAngel generic cao text is only in debug mode 14:55 RealBadAngel in this pr they asked for a way to name an object 14:56 RealBadAngel nametags does it 14:56 est31 no 14:56 est31 nametags is a different system 14:56 est31 the issue is about a different thing 14:56 RealBadAngel i know 14:56 est31 the issue title 14:56 RealBadAngel ok, slightly different 14:56 est31 "Ability to change hover text" 14:56 est31 hover text 14:56 RealBadAngel infotext 14:57 RealBadAngel you know that you can set it, right? 14:57 est31 RealBadAngel, I dont change an issue so that I can close it 14:57 RealBadAngel thx to that you see name of dropped item for example 14:58 RealBadAngel so if you want the spinning dirt to be named "cow: pregnant" here you go 14:59 kahrl on a semi-related note, is the automatic debug infotext actually useful to anyone? 14:59 kahrl if not, I would suggest to remove it 15:00 RealBadAngel https://github.com/minetest/minetest/commit/87291ea44a61614d5bd9efc0657926da3f867b5f 15:00 RealBadAngel that allowed to set infotexts 15:00 kahrl it makes minetest seem less polished IMHO when compiled in debug mode (which quite a number of people seem to do) 15:01 RealBadAngel est31, so either he meant nametag or infotext both are possible already 15:01 kahrl same with the default debug f5 mode, but changing that might be more controversial 15:02 RealBadAngel kahrl, i find that more annoying than useful, so i agree 15:03 est31 kahrl, release builds are not defaulting to f5 mode 15:03 est31 only debug builds 15:03 kahrl est31: I know 15:03 kahrl but lots of people use debug builds anyhow 15:03 RealBadAngel wonder what for? to slow down the game? :) 15:04 est31 because we dont release often enough I guess 15:04 est31 last release is a half year ago 15:04 kahrl http://irc.minetest.ru/minetest-dev/2016-02-15#i_4534171 15:04 RealBadAngel we do have shitload of new code, lotsa fixes 15:05 est31 yea seen it 15:05 RealBadAngel new release would be nice. and is expected 15:05 est31 if somebody wants to be release manager, fine with me as well 15:05 est31 any day 15:05 est31 just I wont be the release manager :) 15:08 RealBadAngel ha! 15:08 RealBadAngel found one issue that my pr fixes :) 15:09 RealBadAngel #1404 15:09 ShadowBot https://github.com/minetest/minetest/issues/1404 -- water opacity bug with irrlicht 1.8.x 15:09 RealBadAngel look at the date :) 15:09 RealBadAngel and in this one you have a list of related ones :) 15:09 est31 they are mostly dupes 15:10 RealBadAngel propaby yeah 15:10 kahrl now it makes sense why it would depend on irrlicht version :) 15:11 Fixer https://cloud.githubusercontent.com/assets/3192173/4872765/0783bbde-61f2-11e4-88b8-0920c4edeabb.png this water bug is of different nature 15:11 est31 RealBadAngel, you were right about #2516 sorry about that 15:11 ShadowBot https://github.com/minetest/minetest/issues/2516 -- Ability to change hover text for objects and mobs 15:11 est31 I have mixed it with some other change 15:11 est31 by blockmen 15:11 kahrl Fixer: yep 15:12 kahrl Fixer: that's missing transparency sorting 15:12 RealBadAngel est31, i may be known of being full mouth, but lately i rather try to code first then talk ;) 15:13 RealBadAngel also when im not sure about something i would rather shut up 15:19 est31 RealBadAngel, sorry for not reviewing every single commit thoroughly. 15:20 est31 RealBadAngel, when you referenced the commit I have seen that I was wrong 15:20 est31 hrmm I seem to have terrible mesh gen lag 15:20 est31 collisions already working and selection boxes 15:20 est31 but not showing up in world 15:20 est31 classical meshgen problem 15:20 RealBadAngel est31, do you have latest commit in? 15:21 RealBadAngel https://github.com/minetest/minetest/commit/7ea40e45b1bf893aa693c179cbe8612ede86faee 15:21 RealBadAngel that speeds up shaders one 15:23 Fixer est31, lag when placing/digging blocks? you should update to latest git 15:24 est31 Fixer, RealBadAngel yup I have that one 15:24 est31 lemme see whether I can reproduce it in minimal development test 15:25 RealBadAngel ok 15:25 Fixer lol 15:25 Fixer crap 15:25 RealBadAngel est31, im balancing now between making both cases fast 15:25 RealBadAngel i believe that last one shall be good enough 15:27 Fixer PR 3695 fixed lag problem for me, but I tested before some changes to it 15:28 est31 hrmmm it really is reproducible 15:29 RealBadAngel est31, first of all, shaders or non shaders? 15:29 est31 mesh lag of multiple seconds 15:29 est31 I have shaders 15:29 RealBadAngel seconds?? 15:29 est31 yes, when I dig fence i can walk into it 15:30 est31 and when I go to unloaded area I can walk around in nothing 15:30 est31 RealBadAngel, tone mapping disabled, bumpmapping enabled, parralax enabled, waving stuff disabled 15:31 Fixer o.O 15:31 Fixer i will compile too 15:35 est31 I'm on singleplayer with technic mod 15:35 est31 but server lag should not influence meshgen 15:35 est31 if collision stuff works the mapblock is on the client already 15:37 kilbith for the next release manager, paramat could take this role as he actively supervises 2 repos and ensures a good tracking for issues 15:37 est31 he is comfortable with mtgame yeah 15:38 RealBadAngel est31, bump and relief are the only cases to enable tangent space 15:38 est31 and he doesnt reject attempts from the release manager to fix bugs in mtgame 15:38 RealBadAngel please do check if you have lags without those 2 options enabled 15:39 RealBadAngel theres one costly thing run recalculate normals (and tangents and binormals) 15:39 RealBadAngel but that we had before for all the cases 15:39 est31 as long as he doesnt start "adjusting" default values in minetest to "reduce processing" because the "release has to be fast", I'm okay with paramat. 15:40 RealBadAngel so maybe is that all other cases have speeded up you call the other slow 15:40 kilbith nobody is perfect 15:41 est31 RealBadAngel, no speed up even with shaders disabled 15:41 RealBadAngel if this method will turn up too slow i will write my own vector calculations 15:41 est31 I'll restart 15:41 est31 some settings can be static 15:42 est31 nope nothing after restart either 15:42 RealBadAngel i will try to profile it 15:42 RealBadAngel it shall be faster no matter what 15:43 RealBadAngel only thing come to my mind are external calls to irr 15:44 RealBadAngel Fixer, have you tried the current master? 15:44 RealBadAngel kahrl asked for some changes before PR went into master 15:48 RealBadAngel est31, is gameplay smooth or it is stuttering for you? 15:50 est31 as in? 15:50 RealBadAngel it may be related, ive also experiencing it lately for a short period of time then it was mysteriously gone 15:50 RealBadAngel choking movement for example 15:50 RealBadAngel as higher peakk 15:51 RealBadAngel that maybe in fact related to meshgen at all, some folks are reporting it for long time already 15:51 Fixer RealBadAngel, compilation near 100% 15:51 RealBadAngel but thats not a new problem 15:51 RealBadAngel it just seems to be popping up randomly 15:51 est31 I dont really have stuttering gameplay 15:52 RealBadAngel cpu? 15:52 est31 idk 15:52 Fixer est31, i was proposing to generate meshes more slowly per rendered frame, you said there is a queue, but it needs improvement, right? 15:52 RealBadAngel you have rather fast one? 15:52 est31 RealBadAngel, sorta yes 15:52 RealBadAngel it may be more visible on slower boxes 15:52 RealBadAngel so game jams 15:53 RealBadAngel i will examine whole meshgen for time spent there 15:53 est31 RealBadAngel, I see the CPU go up during meshgen 15:53 RealBadAngel it cant be like that im making it faster and it chokes 15:53 RealBadAngel theres something real wrong in here 15:55 Fixer RealBadAngel, it is slow again, testing more 15:57 RealBadAngel kahrl insisted on adding reserve 15:57 Fixer RealBadAngel, hold on 15:57 kahrl RealBadAngel: no, that wasn't me :P 15:57 RealBadAngel oops? 15:57 Fixer RealBadAngel, wait, maybe i'm wrong let me look more 15:57 RealBadAngel somebody did :) sorry kahrl :) 15:58 kahrl but reserve() shouldn't cause meshgen slowdown anyway 15:58 RealBadAngel just trying to find a difference 15:58 RealBadAngel est31 says its common for shaders/non shaders, so recalculating normals is not the reason 15:59 Fixer est31, normals enabled or disabled in bumpmap? 15:59 RealBadAngel it will be something equally obvious as textnode for water transparency i bet 16:00 Fixer est31, post your config 16:00 Fixer RealBadAngel, in singleplayer i don't have any problems 16:00 RealBadAngel it may be broken drawtype or something like that 16:00 Fixer RealBadAngel, let me recheck Just Test multiplayer, maybe that was server lag 16:01 est31 Fixer, I can reproduce with shaders disabled 16:01 Fixer est31, no problems with shaders disabled 16:01 RealBadAngel lol 16:01 Fixer est31, are you sure you are at minetest-0.4.13-7ea40e4-win64 ? 16:01 est31 Fixer, I have no win :) 16:01 est31 otherwise probably yes 16:01 Fixer 7ea40e4 16:01 Fixer probably? 16:01 est31 latest master commit 16:02 Fixer i have no problems with shaders disabled 16:02 Fixer hmhm 16:02 RealBadAngel guys, have to go now, cooking lunch, be back in an hour (but will be reading from time to time) 16:02 est31 happy eating! 16:03 Fixer est31, post your config 16:06 RealBadAngel est31, try to disable mapblock_mesh.cpp line 1109 - all the extra drawtypes, see if it changes anything 16:07 RealBadAngel you shall have there only solid nodes 16:07 est31 RealBadAngel, I'm bisecting now 16:08 Fixer lol, when place some block my draw speed makes spike from 0.01 to 0.03 and back to 0.01 16:08 Fixer or even to 0.04 16:10 Fixer est31, still no regression for me, only usual meshgen stutter :( 16:11 Fixer smooth curve at 0.01 s and big bump after placing just 1 block 16:22 Fixer https://i.imgur.com/erdbo5l.gif gif illustration of stutter when placing one block 16:25 Fixer in ultra quite multiplayer environment 16:32 est31 okay RealBadAngel the faulty commit seems to be really the last one 16:33 est31 7ea40e45b1bf893aa6 has introduced the slowness 16:35 RealBadAngel est31, kinda impossible :( 16:35 RealBadAngel it replaced way slower method to get tangent space vertices 16:36 RealBadAngel without shaders even you have exact same method even as used way before, the one thx to which paramat has gained back 20% fps 16:37 RealBadAngel there has to be some stupid bug that is causing unexpected slowdowns 16:38 RealBadAngel i need to force that stuttering to pop up once for all to eliminate it 16:39 RealBadAngel theres no logical explanation why same methods are once faster for some, slower for others 16:40 RealBadAngel and i afraid its deeply burried there, because the meshgen lag is not a new issue, it pops up from time to time 16:43 RealBadAngel and i wont be suprised if it will be something related to minimap and threads 16:45 est31 hah 16:45 est31 you still would like it at 100% wont you 16:45 est31 man I miss these old discussions 16:45 est31 "on intel its only 70%" 16:45 RealBadAngel just a guess ;) 16:46 RealBadAngel but i admit it needs optimization 16:46 RealBadAngel lets try to disable it and see how it affects meshgen lag, ok? 16:48 Fixer disable what? 16:49 RealBadAngel minimap 16:49 Fixer ah, no meshgen lag for me, it was fixed, i was talking about stutter, it has different nature 16:49 RealBadAngel it seems to be connected 16:49 RealBadAngel something is stopping meshgen 16:50 RealBadAngel i can see it on minimap actually, new blocks are appearing there at very slow ratio 16:50 RealBadAngel so, meshgen is slow 16:50 Fixer RealBadAngel, sometimes when i walk in singleplayer it seems 3-4-5 meshes generated, which is fine, but sometimes HUUUGE 50 meshes thing popups %) 16:50 Fixer thing=spike 16:51 RealBadAngel i can see that too 16:51 RealBadAngel they come in packets 16:51 Fixer yes 16:51 Fixer my suggestion was to limit number of meshes generated per rendered frame 16:51 RealBadAngel theyre not prepared in main thread, neither per frame 16:51 Fixer so instead of 50 one you will get 2-2-2-2 over big period 16:52 Fixer just spread them out somehow 16:52 RealBadAngel once mesh is ready, updating it is fast 16:52 RealBadAngel something is holding them from being rdy 16:54 est31 RealBadAngel, client.cpp line 177 16:54 est31 ScopeProfiler sp(g_profiler, "Client: Mesh making"); 16:54 est31 change that to 16:54 est31 ScopeProfiler sp(g_profiler, "Client: Mesh making", SPT_GRAPH_ADD); 16:54 est31 then press f5 to see the graphs 16:54 RealBadAngel est31, thats not minimap slowing it 16:54 RealBadAngel just checked 16:57 RealBadAngel est31, what are numbers there? 16:58 RealBadAngel 0.6 - 2.3 16:59 est31 one number is the average 17:00 est31 the other is I think the "last" one 17:00 est31 idk 17:00 est31 doesnt really use that subsystem 17:00 RealBadAngel they seem to be the same for tangent/regular vertices 17:00 est31 but for before your patch I notice the value going up really high 17:00 RealBadAngel and the lag is here in both cases 17:01 RealBadAngel disabling now special drawtypes 17:02 RealBadAngel the value is now 0.022 17:02 RealBadAngel no lags at all 17:02 est31 yeah 17:02 est31 thats fine 17:02 est31 I get values up to 4 17:03 RealBadAngel lemme bring the minimap back in 17:04 RealBadAngel 0.040 17:05 RealBadAngel so, its special drawtypes that slowing it so much 17:06 RealBadAngel going to disable drawtypes one by one now 17:11 Fixer another classic example of stutters (not related to est31 thing probably): https://imgur.com/8Zz4woj (in gifv) 17:11 RealBadAngel btw, can see that all the time: 17:11 RealBadAngel 2016-02-15 18:09:36: WARNING[Main]: Map::getNodeMetadata(): Block not found 17:11 RealBadAngel 2016-02-15 18:09:46: WARNING[Main]: Map::getNodeMetadata(): Block not found 17:11 RealBadAngel 2016-02-15 18:10:00: WARNING[Main]: Map::getNodeMetadata(): Block not found 17:12 Fixer RealBadAngel, seen those warnings too 17:12 Fixer have no idea why 17:12 Fixer RealBadAngel, week ago or more 17:17 RealBadAngel ok we have a winner i think 17:17 RealBadAngel those are leaves 17:18 est31 leaves? 17:18 est31 that can be it 17:20 RealBadAngel est31, please change leaves style to simple 17:20 RealBadAngel and see if lag is still here 17:20 RealBadAngel that will change drawtype to glasslike 17:21 est31 yup that's it 17:22 est31 with simple leaves the meshgen lag is gone 17:22 RealBadAngel now lets find out whats wrong with allfaces 17:26 RealBadAngel cant see anything wrong there 17:26 RealBadAngel but, we had issues with leaves before didnt we? 17:26 RealBadAngel premesh buffers and indices 17:28 est31 RealBadAngel, created bug #3714 https://github.com/minetest/minetest/issues/3714 17:28 ShadowBot https://github.com/minetest/minetest/issues/3714 -- Meshgen slowdown regression with fancy leaves 17:29 RealBadAngel ok 17:29 RealBadAngel but we dont know yet if allfaces is the reason itself 17:30 RealBadAngel im checking now premesh buffers 17:34 RealBadAngel ok, theres something interesting in here 17:35 RealBadAngel looks like from time to time meshgen loops itself 17:35 RealBadAngel est31, find first meshcollector append 17:35 RealBadAngel at the end of mapblock_mesh.cpp 17:36 RealBadAngel https://github.com/minetest/minetest/blob/master/src/mapblock_mesh.cpp#L1411 17:36 RealBadAngel and put dstream< so more vertices 18:39 RealBadAngel back to buffers then 18:40 Fixer .__. 19:05 kaadmy Obani: the main reason i put that there was to prevent people from copying Pixture and distributing it without any credit 19:05 kaadmy mainly on the Google play store 19:16 Obani kaadmy, well, I think a GPLv3 would be enough 19:16 Obani But I admit, with that definition of ©, that it is a simpler way to tell it, and to make it understand 19:16 Obani able 19:59 ShadowNinja #3347 look good? 19:59 ShadowBot https://github.com/minetest/minetest/issues/3347 -- Fix jumping at node edge by ShadowNinja