Time Nick Message 01:48 Miner_48er kaeza any additional info you need to help with that xban2 issue? 05:42 ShadowNinja ''tell sapier That bug isn't reproducable AFAIK, send me any debug info you can get on it, but it may be beyond my understanding. Also, I'd suggest using distance instead of time for the movement prediction patch to make it more precise. 05:42 HLuaBot I'll tell that to "sapier" next time I see them around. 08:50 Calinou how sane is it to use “dedicated_server_step = 0.01"? 08:53 celeron55 if you want 100% CPU consumption, i guess that's quite effective 08:55 Calinou https://forum.minetest.net/viewtopic.php?f=3&t=9766&p=148708#p148684 08:55 Calinou this guy's server seems to defy logic then 08:57 celeron55 it depends on player count of course 11:09 Megaf Minetest is segfaulting so often now a days that we have a server named Meowtest "SEGFAULTY" 11:10 Megaf and as we know, segfault means code error 11:11 Megaf 19:05:24: ACTION[ServerThread]: FireStar places node wool:red at (479,16,56) 11:11 Megaf Segmentation fault 11:14 VanessaE ... 13:42 RealBadAngel https://www.youtube.com/watch?v=ixgAxnBTIi4 13:42 RealBadAngel any objections on adding param2 to plantlike? 14:09 RealBadAngel PilzAdam, https://www.youtube.com/watch?v=ixgAxnBTIi4 14:10 RealBadAngel how do you like it? 14:18 PilzAdam why param2? and why are the plants flipping when being placed? 14:19 RealBadAngel param2 because its made for things like that, and flipping because i just modified grass def, with after_place 14:20 PilzAdam what are "things like that"? 14:20 RealBadAngel modifiers 14:20 Zefram_Fysh in this case controlling the rotation? 14:20 RealBadAngel its just additional variable we can use 14:20 RealBadAngel code for that is dead simple 14:20 RealBadAngel 2 lines in core modified 14:25 RealBadAngel https://github.com/minetest/minetest/blob/master/src/content_mapblock.cpp#L1119 ---> http://pastebin.com/m9jdAdrY 14:41 PenguinDad PilzAdam: #802 seems to be fixed 16:19 Krock again 16:19 Krock - Krock quit (*.net *.split) 16:19 Krock wtf 16:39 paramat RealBadAngel, i made this issue to remind you ;) https://github.com/minetest/minetest/issues/1462 17:04 sapier ShadowNinja: Time is already present in engine, distance would've to be calculated separate and as I mentioned in discussion on minetest would introduce quite some new ways to get inconsistent states. 17:04 sapier For a bug not beeing reproduceable I had it quite often yesterday .. guess about 40 times 17:04 sapier within only a few hours 17:05 sapier I wasn't able to exactly locate it but it somehow seems to be related to punching and dieing 17:06 sapier another important thing may be that there's no player involved 17:13 sapier megaf could you try to find out what is done to do the segfault? or run in debugger so we get a backtrace 17:18 Megaf sapier; it happens miliseconds after you start minetest 17:18 Megaf or right after you put your password and press login 17:22 sapier no it doesn't 17:23 sapier just tried didn't happen (for me) 17:32 paramat Here are 2 proposals for improving my wieldhand texture https://github.com/minetest/minetest_game/issues/266 17:44 Calinou “mgv6_freq_desert = 0.4” is lower more deserts or less?! 17:44 Calinou I want a world with a lot of beaches, but less deserts 18:01 RealBadAngel bump: https://github.com/minetest/minetest/blob/master/src/content_mapblock.cpp#L1119 ---> http://pastebin.com/m9jdAdrY 18:02 RealBadAngel https://www.youtube.com/watch?v=ixgAxnBTIi4 18:02 RealBadAngel any objections? :) 18:03 sapier I don't like "this video isn't available" ;-) 18:04 Calinou mgv6_np_beach = 0, 1, (250, 250, 250), 59420, 3, 0.5 → WTF does each setting mean? 18:04 Calinou many values have a similar formatting 18:04 RealBadAngel sapier, ? 18:05 sapier hmm ok youtube isn't kompatible to conqueror only 18:05 sapier what exactly is the change? 18:06 RealBadAngel make plantlike twisted by holding additional angle in param2 18:06 RealBadAngel 0 - no change, current behaviour 18:06 sapier what's param2 usually used for? 18:06 RealBadAngel many things 18:07 RealBadAngel level, 6dfacedir, wallmounted dir 18:08 sapier so it's not abusing it 18:08 RealBadAngel each drawtype can do with it whatever it wants to 18:08 RealBadAngel plantlike had no use for it before 18:09 sapier well I'd suggest to ask someone else too but I don't see a reason not to add it by now 18:10 RealBadAngel ok 18:19 RealBadAngel and what about https://github.com/minetest/minetest/pull/1487 ? 18:21 sapier RBA I always wonder why you have to mix in behaviour changes to commits telling something different 18:22 sapier e.g. the contrast changes... I don't know if they're good or bad I only no they're irrelevant for speed 18:22 RealBadAngel see the changes 18:22 RealBadAngel theres no sqrt and x*x there anymore 18:23 RealBadAngel takin that it was calculated for each vertex, its a speedup, dont you think? 18:24 sapier sounds even worse you change the behaviour of a function while keeping signature identical ... 18:25 RealBadAngel contrast? 18:26 RealBadAngel it does exactly the same thing as before 18:26 sapier not for what I understand 18:27 sapier sqrt(0.3) != 0.4 18:28 RealBadAngel that idiotic pow and sqrt wasnt my idea, and is completely wrong 18:28 sapier possible so if you do understand what's done there please explain 18:29 sapier right now the only thing I can tell for sure it's not a format change only but really changes value 18:31 RealBadAngel http://irrlicht.sourceforge.net/docu/_s_vertex_manipulator_8h_source.html 18:31 RealBadAngel see line #97 18:32 RealBadAngel this is used only to apply faces shading, so simple contrast change 18:33 RealBadAngel before there was used SRGB to linear color change so color^2 * factor, then sqrt to get modified value back 18:33 sapier So what's the result of that change? for what I see it's for 0.3 -> 0.4 effectively changing 0.55->0.4 18:34 sapier I don't know the range of that value 18:34 RealBadAngel it doesnt really matter if result gonna be slightly different 18:34 sapier if it's 0-100 that change is most likely irrelevant but if it's 0-1 it's more then significant 18:34 RealBadAngel its used only to make top brighter and faces darker 18:35 sapier so we will have more or less contrast? 18:35 RealBadAngel you wont notice a difference 18:35 RealBadAngel just check it out 18:35 sapier where do I have to look at? 18:36 RealBadAngel fire up the game with patch 18:36 RealBadAngel and look around 18:36 RealBadAngel it affects all the nodes you can see 18:37 sapier ok I'll do some screenshots ... hopefully you're right, if not I'll not be amused as it'd be your job to mention significant changes and not silently hide them in a "performance improvement" commit 18:38 RealBadAngel i cant see any difference, sfan5 and VanessaE neither 18:38 sapier possible as I said if range of that value is big enough nothing will happen 18:38 sfan5 if you put before/after side-by-side you can see a small diff 18:38 RealBadAngel propably yes 18:38 sfan5 you won't notice otherwise 18:39 sapier well I don't care about small changes what I don't like is them not beeing mentioned 18:40 sapier because sometimes they aren't as small as they seem to be in some special cases 18:41 VanessaE I certainly didn't notice a change when I tested his patch before, only a very slight performance improvement (marginal, hard to be sure if there was any improvement) 18:41 VanessaE certainly wasn't any *loss* 18:41 RealBadAngel you have fast box 18:41 RealBadAngel for me gain is huge 18:41 RealBadAngel 2x better view distance at least 18:42 VanessaE yeah but Minetest is heavily CPU-bound for me, any improvements in FPS etc are gonna have to happen in whatever parts of its rendering pipeline occur there rather than the GPU 18:42 RealBadAngel i do have 1.8ghz dual core and older radeon 18:42 RealBadAngel this is rater CPU change 18:42 VanessaE ok 18:42 RealBadAngel not GPU 18:43 RealBadAngel when i start to think, nothing in this code makes GPU things faster 18:43 Calinou CPU changes are still good, very useful for gameplay with client + server 18:43 Calinou it's probably more important than GPU right now 18:43 sapier removing that sqrt is a good idea for sure what I still don't understand is why the values had to be changed too ... they're not even close to what'd result on just using the original values 18:43 VanessaE ok then what I saw was legit 18:44 VanessaE sapier: maybe the same reason people often give measures of 0.1 when designing a nodebox, instead of measures of 0.0625. it's easier to write, even if it's horribly inaccurate. 18:44 sapier where does minetest save it's screenshots? 18:44 VanessaE sapier: in the current dir 18:44 Calinou sadly 18:44 VanessaE unless you have screenshot path set 18:44 Calinou should just be ~/.minetest/screenshots by default 18:45 VanessaE screenshot_path = in case you want to fix it somewhere sane,. 18:46 Krock ehm. it saves a white image there when I use F12 18:46 Calinou if it's set, can't take screenshots 18:46 RealBadAngel sapier, i can fine tune the values to be the same as before if you insist 18:46 Calinou oh, directory must exist 18:46 Calinou created screenshots dir by hand 18:47 sapier hmmm for what I see we loose contrast 18:50 RealBadAngel does it really matter when such gfx freak like VanessaE is ok with it? ;) 18:50 VanessaE :P 18:50 sapier yes you should do that RealBadAngel especially as you forgot some values 18:50 sapier It's quite visible for my wood ceiling at night 18:50 VanessaE "gfx freak" he calls me :P 18:50 RealBadAngel ok, i will fine tune the values 18:51 RealBadAngel please review the rest of the code, now im off to work 18:51 sapier old version did show wood texture cleanly while new variant is almost stripes 18:51 VanessaE stripes? 18:51 VanessaE screenshot? 18:52 Calinou fix wood texture then 18:52 Calinou compare using these textures maybe: https://dl.dropboxusercontent.com/u/82342922/minetest/carbone-minetest-pack.zip 18:53 sapier it's default texture and reduction of contrast is not a texture issue 18:54 Calinou the game feels pretty contrasted currently; before 0.4.10, it was way less contrasted 18:54 sapier main difference seems to be downside of nodes is darker while upside seems to be brighter ... http://imgur.com/GRqwqOz,urNcLTW 18:56 VanessaE sapier: well seems like that should easy enough to tweak in his code. mapblock_mesh.cpp lines 1147 and 1149 18:57 VanessaE however it looks like he only changed the bottom, not the top 18:57 sapier for what I see every node changes 18:57 sapier well at least some values haven't been changed at all ;-) 18:58 VanessaE anyway if the bottom is too dark then just raise that value on line 1149 a little further 19:00 PenguinDad but the bottom face shouldn't be brighter than the top face imo 19:00 sapier some of the values are quite close to what they used to be I'll check if I get an exact match if I change it to that value 19:01 VanessaE better change the other figures too then 19:02 VanessaE I guess 0.71 for those 19:21 VanessaE any luck? 19:25 sapier sorry had to do a rescue operation for a sheep ... a real one 19:25 VanessaE oh? 19:26 sapier actually a lamb that went into a tube below a street and didn't get out of there on it's own 19:26 VanessaE wow 19:26 VanessaE glad you were able to help the poor thing. 19:26 sapier stupid animal 19:27 sapier I'm not that sure my mobf animal are really more stupid then real ones ;-) 19:27 VanessaE haha 19:27 Krock mobs do what the programmers say, and if the mobs so stupid things, then ... ehm yes 19:27 sapier http://imgur.com/EqOqG6F,p7LKwLf,lZHDSBE 19:27 Krock *do 19:28 sapier better ... but I still seem to have missed something 19:28 sapier the wall still seems to be more dark then originally 19:29 VanessaE which one of these images is the original code before the patch? 19:30 sapier ahh still wrong value 19:32 sapier the dirty git is the one I fixed, the dark ceiling is rba 19:33 sapier first=original second=rba third=mine 19:34 VanessaE hm, yeah that far wall (with the windows) does look just slightly darker in yours 19:34 sapier I missed some value just trying corrected one 19:34 VanessaE right wall too. but the left wall and ceiling look the same. 19:34 VanessaE oh ok 19:35 VanessaE (compared to the original that is) 19:35 sapier http://imgur.com/Ed4wiSi,OuXex2u,KnGJ0hH#0 that's back to normal, same first=orig ... 19:36 sapier https://gist.github.com/sapier/7ccc51c71a31f3ab984c 19:36 VanessaE I'd call it good. 19:36 VanessaE certainly good enough to merge anyway as far as the lighting is concerned. 19:37 sapier I'd like to know why RealBadAngel changed it at all? Was it just beeing to lazy to calculate it correct or intentionally ;-) 19:37 VanessaE idk, and he's probably gone off to work by now :) 19:37 VanessaE (probably too lazy to calculate it correctly ;) ) 19:37 sapier thought rba is in europe too? 19:38 VanessaE Poland, but he works odd hours 19:38 sapier I see 19:38 sapier well my value are rounded to two digits behind comma too but that seems to be close enough 19:40 sapier conclusion if there's a reason to change it we should talk about it again if not I don't like this to be merged with different values then the original ones 19:41 VanessaE parse error in function sapier.english(). attempted translation: we should not merge at all, and talk about this regardless. :) 19:41 sapier almost :-) 19:42 sapier what I meant is if there's reason to change that contrast numbers I'd like to know them. If there is no reason they should be fixed to result in same contrast as before 19:42 VanessaE in that case, I say go with the numbers you just came up with 19:43 VanessaE if they work for you, they work. period. :) 19:43 VanessaE merge and move on. 19:44 sapier why do you always expect me to merge everything ;-) 19:44 VanessaE because you're good at it? :) 19:45 sapier sounds like a curse 19:45 VanessaE hahahah 19:56 celeron55 what 19:57 celeron55 so you removed my sRGB<->linear space multiplications? 19:57 celeron55 there's a reason for those you know 19:58 celeron55 it's an approximation of the 2.2 (1.8 on macs) gamma used on screens 19:58 sapier you're not talking about the contrast thing aren't you? 19:58 celeron55 which is applied in order to make vertex color multiplications in linear space and then converted back to screen gamma 19:58 celeron55 sqrt is 2, which is close enough to those 19:58 celeron55 yes, i don't understand why it's called contrast 19:58 celeron55 i'm pretty sure i didn't call it that when it wrote it 19:59 sapier still what's the difference in specifying to pass the sqrt value instead of calculating it inside? 19:59 celeron55 ? 19:59 sapier applyContrast second parameter 20:00 sapier mapblock_mesh.cpp 20:00 celeron55 it isn't specified that way, right? 20:01 sapier true but why pass a value just to calculate sqrt from it if the only usage is for compile time known parameters 20:02 sapier sqrt(x*x*y) == sqrt(x*x) * sqrt(y) == x * sqrt(y) 20:03 sapier hope I didn't forget to much about math the last years :-) 20:04 sfan5 sapier: good news ;), MinetestBot just confirmed me that what you said is right 20:05 sapier what exactly? 20:05 sfan5 !c math.sqrt(5*5*7), math.sqrt(5*5) * math.sqrt(7), 5 * math.sqrt(7) 20:05 sfan5 (13.228756555322953, 13.228756555322953, 13.228756555322953) 20:05 sapier ahh ok :-) 20:05 celeron55 yes it really is so, now i'm wondering why i didn't do it that way originally 20:06 celeron55 this doesn't make intuitive sense to me at all and i need to figure out why 20:06 celeron55 (i mean, colorspace-wise) 20:06 celeron55 (math-wise it does) 20:06 sapier ok then I'm not gonna merge it now maybe there's reason we don't understand yet 20:07 VanessaE damn it why does RealBadAngel always have to start this stuff right before work :-/ 20:07 sapier well if he didn't change the values it'd already be merged too ;-) 20:08 celeron55 oh now i understand why it works that way 20:09 celeron55 does anyone in here know anything about colorspaces or should i just skip any explanations? 8) 20:09 sapier I don't have any idea :-) 20:09 VanessaE I only know a little about it. 20:09 celeron55 as long as the values are replaced by the sqrt of the previous value, it's fine 20:09 VanessaE (not enough to speak with any authority on the subject) 20:10 celeron55 don't change them to any kind of decimal approximations, because those will be wrong even if they look like round numbers 20:11 celeron55 and also, of course the difference won't be very apparent, but i believe the wrongness will accumulate and make people feel something is not right while they can't see why they feel so 20:11 celeron55 ...now i'm assuming the previous values made sense to begin with 20:11 sapier http://imgur.com/Ed4wiSi,OuXex2u,KnGJ0hH you did see this? 20:12 sapier First is original second rba's somehow changed values last is aproximated first numbers 20:13 celeron55 hmm, no; the previous values don't make sense in any "realistic lighting" sense 20:13 sapier so we've got values other values and no idea about correct values? :-) 20:13 celeron55 and also i don't know any immediate way to make them so 8) 20:14 sapier ok if we don't know the correct ones imho we should stay at the already used values 20:15 celeron55 so, this has to be either left exactly as it was or a new gut feel has to be taken with each configurable lighting mode 20:15 celeron55 just put the sqrt()s of the old values there 8) 20:15 VanessaE sapier: "already used" values is fine but the problem is getting rid of those sqrt()'s. 20:15 sapier exact with how many digits? :-) 20:15 VanessaE round to 4 digits. 20:15 VanessaE :) 20:16 VanessaE er, 4 decimal places rather. 20:16 celeron55 well given that we are dealing with 8 bit color here, i guess that works fine 20:16 sapier ok I'll recalculate my 2 decimals values and add 4 instead 20:16 celeron55 4 is definitely excessive but that means we are staying true to the original 8-) 20:17 sapier I'll add a comments why there are those strange numbers ;-) 20:17 celeron55 anyway i'm guessing better values could be found by careful testing 20:18 VanessaE I was only kidding when I said 4 digits :P 20:18 celeron55 basically the old values are the multiplication factors in linear color space, and any new values really get their actual meaning from that too (multiplication in sRGB color space doesn't mean anything useful) 20:19 VanessaE with 8 bpp color, 3 digits would give you at worst about 1/4 bit rounding error 20:19 celeron55 and the relation of those is the sqrt 20:19 VanessaE (so maybe 4 digits IS about right) 20:20 sapier I'm gonna do 6 digits ... if someone feels this is to much it can be removed later ... but I don't wanna calculate another time :-) 20:20 VanessaE haha 20:20 celeron55 also i don't know if the shaders actually mess these up further so that the values in fact mean absolutely nothing no matter what is done to them 20:20 VanessaE celeron55: good question. 20:20 celeron55 8D 20:21 celeron55 also didn't RBA change those values already at some point? 20:21 VanessaE er, s/8 bpp color/8 bits ber color channel/ (you know what I meant) 20:21 celeron55 8bpc! 20:21 VanessaE celeron55: yeah, to increase the overall contrast relative to the sides of nodes 20:22 sapier they doo have an effect 20:23 sapier and considering minetest already being to dark at night I don't like it to become even more dark ... but of course that's a subjective opinion only 20:23 celeron55 i mean, depending on what the shaders do the effect could become such that if you set a factor of sqrt(0.5) there, the visual brightness wouldn't be halved like it should but instead just lowers by whatever amount the arbitrary calculations result in 20:24 celeron55 given the amount people probably care about this, this discussion is already useless thoguh 20:24 celeron55 though* 20:24 * celeron55 removes himself -> 20:24 VanessaE celeron55: well for comparison, the original patch that changed the brightnesses is here, https://github.com/minetest/minetest/commit/c9ba92c4ad0b010d6d36890c3b1ddfb2e773560f 20:25 celeron55 oh there my actually useful function name was removed 20:26 celeron55 the old values are arbitrary still though 20:26 VanessaE they were decided kinda by committee 20:27 sapier to be honest I don't understand the exact meaning of those values but changing them without any explanation doesn't seem to be right 20:27 VanessaE calinou, RBA, myself, and at least a couple others, via screenshots RBA was passing around. 20:29 celeron55 i don't think we can construct any kind of useful model for this kind of node shading 20:30 celeron55 or, well, we could, but it would be just as arbitrary 20:30 VanessaE lemme check the logs 20:30 sapier so what's result? stay as close as possible at original values? 20:30 celeron55 because we would have to define an arbitrary material properties for the faces, arbitrary light angle(s) and possibly other things too 20:31 celeron55 stay as close as possible to the revised values by RBA that currently exist 20:31 VanessaE http://irc.minetest.ru/minetest/2014-06-12#i_3757468 20:31 VanessaE there's the bulk of the discussion on the brightnesses. 20:31 celeron55 they have at least some thought put to them even while probably not everyone agrees to them 20:33 sapier https://gist.github.com/sapier/8a13cd2cbe2cc8e6da74 then I'm gonna use this values for merge 20:33 VanessaE sapier: go for it. 21:04 sapier is there any reason to keep default per block entity count at 49 after fixing the duplication issues? 21:06 sapier what about 1024= 1/4 of nodecount or 512 = 1/8 of nodecount ? 21:06 VanessaE not especially so - I run with a default of 500 without issues, as long as there isn't some mod glitching out and creating them like crazy 21:06 VanessaE but bear in mind the usual issue of lower FPS from having lots of entities around 21:07 VanessaE 1024 seems a little excessive, 512 seems about right to me 21:07 VanessaE it can always be adjusted by the server operator if that's still too low 21:08 celeron55 i guess it's starting to be the time to raise the limit 21:09 sapier is there any reason to use pot2 values or would vanessae's tested 500 value be good too? 22:28 celeron55 sapier: i personally never use 2^n values unless specifically needed