Time Nick Message 00:39 VanessaE although GPU time is more precious, remember too that there are things a GPU can do that a CPU might be 100x slower at doing. 00:39 VanessaE better to spend GPU cycles if the result is faster and doesn't hinder the ability to use the GPU for other things 02:56 RealBadAngel VanessaE, finalColorBlend is implemented both sides 02:57 VanessaE I was speaking in general terms. 02:58 VanessaE however in this case, which one is faster/ 02:58 VanessaE ? 02:58 RealBadAngel and i think CPU's one is faster, mainly because its called once, when ratio changes 02:58 VanessaE then do it in the CPU, by all means. 02:58 RealBadAngel while on GPU each render pass 15:36 nore sfan5, BlockMen: what about improving ore generation in minetest_game, using the noise_threshhold option (so the ores don't spawn everywhere, you have to find an ore mine) to make mining more interesting? 15:37 sfan5 sounds interesting 15:51 Sokomine i'm not sure if that's such a good idea. it might work on servers where there's much traffic and lots of players for trading. in general, the "everything is everywhere" approach works quite well 15:53 nore well, the scale should not be too high 15:53 nore perhaps 100 nodes 16:17 nore sfan5, http://nore.mesecons.net/screenshot_3665078645.png 16:17 nore (made the ores generate in air for testing) 16:18 nore thoughts? 16:19 proller COMMIT! 16:23 VanessaE_ nore: holy moses 16:23 nore hm? 16:23 VanessaE_ that's a bit too scattered imho 16:24 VanessaE_ make the groups larger, and more spread apart 16:24 VanessaE_ or wait 16:24 VanessaE_ that's the default pattern isn't it 16:24 nore it's the default (but more dense) where the noise is high enough 16:24 VanessaE_ hm 16:25 nore for mese, you'll have to find an area where it spawns, and then search around it 16:25 VanessaE_ I stand by my previous statement then :)_ 16:26 nore for ores such as coal or iron, once you've found an area, there are a lot of other ones around 16:26 VanessaE_ raise the threshold and make the spawning more dense - so that the ores group together more. create more dead zones between ores 16:27 nore it is there: https://github.com/Novatux/minetest_game/tree/new_ore_gen 16:27 nore ^ no, that's a problem 16:27 nore it becomes too easy to get a lot of ores when you've found an area 16:28 VanessaE_ well just raise the threshold then and keep the density the same 16:28 nore of which ones? 16:28 VanessaE_ just iron and coal, by the look of it. 16:28 nore actually, in the screenshot, there is a *huge* area without ores 16:28 VanessaE_ or is that copper on the left? 16:28 VanessaE_ I can't tell 16:28 nore copper 16:29 VanessaE_ oh ok 16:29 VanessaE_ from this screenshot it's kinda hard to tell what's where 16:29 VanessaE_ the mese and diamond look about right though 16:29 nore well, I see no iron in it 16:29 VanessaE_ maybe a bit more dense on diamond, if what's there is average 16:29 nore it's hard to get all the ores at once 16:30 nore there's some gold on the left 16:30 VanessaE_ I see it. 16:30 nore ah, and iron too 16:30 VanessaE_ american cheese ore. 16:31 nore well, you can try it if you want 16:31 VanessaE_ naw 16:31 VanessaE_ I'll leave it in your capable hands 16:31 nore problem of screenshots like that is that it is difficult to see how hard it is to get those ores when they're in stone 16:32 eugd now how about tying this to biomes? 16:33 nore hm.. not all mapgens support biome 16:35 nore btw: the iron/copper/coal may look close, but it's still not trivial to find them when you've got one 16:36 nore and for other ores it's more like giving you some vague hint about where to dig 16:39 nore there, #301 16:43 VanessaE_ https://github.com/VanessaE/dreambuilder_game/commit/1a76e272727de63a2f4237fb438f52925b3b234d :-) 16:43 nore :) 17:07 sapier anyone to comment this http://dev.minetest.net/Lua_code_style_guidelines? 17:11 Calinou in before this violates every single of my own guideline 17:11 rubenwardy It looks okay 17:11 Calinou foo({baz=true}) 17:11 Calinou ugly 17:12 Calinou foo({baz = true}) 17:12 Calinou is way better 17:12 rubenwardy ^ +! 17:12 rubenwardy * +1 17:12 VanessaE_ wtf? 17:12 sapier that's allowed by the guidelines ... but it's not written very obvious ... I interpreted it wrong too 17:12 VanessaE_ what's with all the ^^^---- ------vvv crap? 17:12 sapier those are my comments 17:13 sapier wiki isn't a good env to do discussions 17:13 Calinou someone who didn't learn wikicode 17:13 Calinou there's a talk page. 17:13 rubenwardy But you said comments should be done with -- Capital 17:13 Calinou -- This is how I write comments, as full sentences. 17:13 rubenwardy also works in wikis 17:13 Calinou lots of people hate on this, but it makes the most sense 17:13 Calinou rubenwardy, any HTML does ;) 17:13 sapier yes but talk pages are not WITHIN the page but separate ... for reviewing that's quite silly 17:15 sapier it's temporary to mark thinks I consider to be at least questionable 17:15 Calinou use italic text? 17:16 sapier and how to mark the part I'm talking about? 17:16 rubenwardy ~~!~~ LOOK TO THE PLACE THE IS ABOVE THIS TEXT ~~!~~ 17:16 Calinou ↑ ↓ 17:17 Calinou ↙↓↘←↔→↖↑↗↕↔ full set :D 17:17 sapier and what's the difference to the ^^^markers? ;-) 17:17 VanessaE_ what, no four-way arrow? ;) 17:18 sapier well I tried italic right now ... the difference is almost non existent on my font ... wiki is just wrong tool for doing reviews :) 17:19 sapier I suggest discussion those things fast to be able to remove the markers quick ;) 17:19 Calinou they look proper 17:19 Calinou v is a letter, ^ is exponent 17:20 sapier do we need to write a style guide to define how do discuss about style guides first ? 17:24 sapier well some things actually aren't even worth discussing as this is minetest core styleguide and minetest does support lua 5.1 as base version only there ain't any discussion about 5.2 functions 17:24 sapier at least until minetest upgrades to 5.2 17:27 sapier >>Don't compare values explicitly to true, false, or nil, unless it's really needed.<< either disallow it or skip unspecific rules like that one at all 17:28 sapier especially as that specific example is plain wrong, comparing for nil instead of false IS a valid check 17:47 sapier http://dev.minetest.net/Lua_code_style_guidelines do you feel better this way of commenting? ;-) 17:50 Krock yay! guidelines 17:50 PenguinDad I'm not too sure about the hh############################################################# lines 17:51 sapier they're gonna be removed once it's no longer a draft ;-) 17:54 celeron55 can someone explain what all those ^------- things are 17:54 sapier comments ? 17:54 celeron55 why don't they have any mention about who made the comments 17:55 sfan5 maybe we should use an etherpad like we used it for minetest.features 17:59 celeron55 a wiki is fine for commenting on things, as long as people comment so that the comments are organized properly (for example, they should contain the name of the author of the comment) 18:00 sapier it's not even close to as comforable as commenting code on github ;) 18:01 sapier but I fixed my comments anyway 18:03 kahrl "Paranthesis always improve readability a lot" ok, let's write mods in lisp :P 18:04 sfan5 :P 18:04 sapier https://github.com/minetest/minetest/pull/1557 can someone test this? I know modders don't like to be told "your mod consumes way to much time per step" but it's gonna help a lot 18:04 sapier well kahrl I wrote my comments mimicrying level of coding style 18:05 sapier somehow right but completely wrong for some other things ;-) 18:05 kahrl heh 18:06 sapier yet most of the guidelines are fine except thos parts ShadowNinja wrote to have something to complain about ;) 18:18 Krock Ehm.. the serverlist hasn't been fixed yet. is it going to happen client- or serverside? 18:18 Krock (talking about *CDP display) 18:27 kahrl Krock: issue #? 18:28 Krock kahrl, pull of 1 month ago (feeling) 18:28 Krock +request 18:29 Krock ah right. I needed to find someone who agrees that merge 18:29 VanessaE_ https://github.com/minetest/minetest/pull/1480 18:29 VanessaE_ kahrl: ^^^^ 18:30 sapier krock why didn't you squash it? 18:30 kahrl sapier: eh, the one who merges it can squash it 18:31 Krock and I can't merge.. 18:32 sapier of course... as the one who merges it can do the code style fixing the bug fixing ... code writing ... but yes in this special case it's not a lot of work 18:32 kahrl what I find more problematic is that 1. the C++ side passes parsed json to lua unchecked (not problematic in itself) and 2. the lua side doesn't do any checks at all 18:33 sapier nice to know I already mentioned that and I have been punched for this opinion 18:34 kahrl have you? 18:34 Krock well, if the client can not use it, it will ignore it.. 18:34 sapier discussion? ;-) 18:35 kahrl sapier: I don't see it 18:35 sapier "not sure if it's a problem or an additional feature but this code would cause ANY element sent to be transfered to lua not those we support only" 18:36 kahrl sapier: yeah I saw that, but I didn't see anyone bashing you for it 18:36 sapier well that's been here in minetest-dev 18:37 sapier still I stay at my position so at least I will not agree to it ... so whoever merges it needs two other agreements 18:38 kahrl I would agree to b761d2d at least 18:39 kahrl the rest I only agree to if the lua side validates the data 18:39 sapier lua side validating data still violates rule to catch errors and undefined behaviour as soon as possible 18:40 sapier if there's some malicious data in server list response passing it to lua opens up another huge part of software containing errors that can be exploited 18:41 sapier that's why I don't agree to just pass everything 18:43 kahrl so let's merge b761d2d? 18:44 sapier b761d2d seems to be fine for me too 18:45 kahrl I'll merge it in an hour then 18:49 Krock fine. 18:54 kahrl btw: this would've been far harder had the commits already been squashed :P 18:55 sapier well if you want it that way I can stop squashing all of my pulls ;-P 18:55 kahrl :P 18:57 kahrl well what I mean is, squash commits that belong to the same change, but don't squash commits that do different things 18:57 kahrl fixing the *CDP display and switching to push_json_value are different things 18:57 sapier if course but whatever was changed in the commit you're gonna merge was reverted by later changes 18:58 kahrl true 18:58 sapier in this case it's actually history not different things ... for this special case history is good ... usually I don't wanna have the early experimental attempts ;-) 18:59 kahrl I guess the way to go is: if you see devs fighting over a later change, don't squash those later changes with earlier changes 19:01 sapier yes but I guess we won't have something similar to this case soon ;-) 19:01 kahrl no fighting between devs? you sure? 19:01 sapier that part is permanent ;-P 19:01 sapier the previous and later changes thing ;-) 19:02 kahrl yeah 19:02 sapier btw who feels responsible for modstore? 19:02 kahrl though it happens from time to time 19:02 sapier to be more precise mmdb 19:02 kahrl not me :P 19:02 sapier seems like it's broken atm 19:03 Calinou fixing weblate is more important 19:04 sapier both is important 19:04 sapier while mmdb results in direct user experience 19:05 sapier maybe we should replace it by some sort of linklist in minetest github repository combined with a github interface 19:06 sapier mod list containing links on minetest and some sort of description to be placed in the mods own github repository (or any other git repository) .. or other vcs if someone implements the interface 19:25 sapier https://gist.github.com/sapier/3b6b8327d1fefadd72cc Any comment to implementing mod store this way? 19:26 Krock sapier, how about a forum link? 19:27 sapier type forum would be ok too but it's way more hard to parse a forum post then download files from some webfrontend 19:27 * Krock likes that idea 19:28 sapier well I'll not implement the forum thingy as I believe it not to be any more reliable then current variant 19:28 sapier but as there's a type in my specification the design is open for it 19:31 kahrl sapier: what about mods with rolling release? 19:31 kahrl i.e. no version numbers 19:32 sapier special "tag" "LATEST"? 19:32 kahrl could work 19:32 kahrl or HEAD to keep with git terminology 19:32 sapier yes 19:33 sapier we couln't do voting ... but that feature didn't even work by now so not a big loss 19:34 eugd do mapsectors have metadata? 19:34 sapier the only bad thing would be as it's lua code we'd rely on content beeing correct and can't verify content in c++ easyly 19:35 eugd i see functions referencing that, but also comments that it's depreciated? 19:35 sapier can you provide an example? 19:35 eugd looking in map.cpp and map.h 19:35 eugd map.cpp has functions dealing with mapsector meta 19:36 eugd and map.h mentions the possible depreciation 19:36 kahrl sapier: if you remove the "return" and replace = with : it'd be json 19:36 * kahrl hides 19:37 sapier as we already have a json parser that be a possibility ... yet the other way would be way more easy 19:38 sapier the only difference to serverlist is modlist isn't automatically loaded each logon ... not sure if this is enough to allow direct passing to lua 19:38 sapier -logon +start 19:39 kahrl wait, where are these file stored and who executes them when? 19:40 kahrl files* 19:40 sapier first one in some repository on minetest github account 19:40 sapier last one within the mod itself 19:41 kahrl I don't think downloading lua code and executing it (before you accept to download a mod) is a great idea 19:41 sapier eugd seems to be almost completely dead code, the only thing written to metadata is a version number 19:42 sapier kahrl: you're right 19:42 sapier so json 19:42 sapier uargh 19:42 kahrl hehe 19:43 kahrl we could hack libjson to parse lua table syntax (without executing it as lua) instead ;) 19:50 sapier https://gist.github.com/sapier/3b6b8327d1fefadd72cc vetter? 19:51 kahrl sapier: looks good 19:51 kahrl oh, line 5 needs updating 19:52 sapier done .. I'm not sure if last one is correct json anyway but it's good enough to show how it's intended 19:52 kahrl yeah 19:53 kahrl sapier: hmm, there could be one problem 19:53 sapier yes? 19:53 kahrl validated_versions lists tags but any modder could force push to replace the tags 19:53 kahrl maybe list the corresponding commit hashes as well? 19:54 sapier would limit us to github ... maybe md5/sha sums? 19:54 sapier of zip file 19:54 kahrl yeah I was about to suggest that 19:56 sapier I'd suggest support of md5 and sha256 19:56 kahrl even just sha256 19:56 kahrl md5 shouldn't really be used anymore 19:56 Megaf why not? 19:57 sapier hmm guess current hw will be fast enough but we'd need to implement it ourselfs 19:57 Megaf its so good to check file integrity 19:57 Megaf but it is slow 19:57 kahrl Megaf: there have been a number of attacks 19:58 sapier yes kahrl but I'd guess we remove a mallicious mod from our list too 19:58 kahrl the first one in 1996 19:58 kahrl sapier: still, no reason to use an outdated hash algorithm 19:58 sapier well md5 is easy to implement sha256 too? 19:59 sapier we're not writing high security code ;-) 19:59 kahrl I guess there are premade implementation 19:59 kahrl +s 19:59 sapier lglp? 19:59 kahrl public domain probably 20:00 kahrl although sha256 is designed by the N*A... maybe use something else? :) 20:01 sapier did you just volonteer to implement and verify and proofe a brand new hash algorithm? :-) 20:01 Calinou http://jxself.org/skein.shtml GPL though 20:01 Calinou ask author for exception? 20:01 Calinou there may be other implementations of it 20:02 kahrl whirlpool? 20:02 kahrl that one is public domain 20:02 sapier we should use something ppl can verify too not only we 20:05 kahrl we could allow ./minetest --whirlpool with only a few lines of code 20:05 kahrl or whatever we end up choosing 20:05 sapier yes but I'd still prefer standard tools 20:06 kahrl most people are using windows, they are out of luck anyway 20:06 kahrl on linux you can install md5deep 20:06 kahrl (which implements whirlpool too) 20:07 sapier still fearing nsa having a backdoor to sha256 for a program having security level like minetest is like replacing heavy reinforced wooden front door by steel door while having a paper backdoor 20:08 kahrl I guess you're right 20:09 kahrl we should still implement ./minetest --sha256 for those poor windows users 20:10 sapier if they install a mod from store check is done automatically and if we manage to fix the localization issues we can reenable the in game install of local mods too 20:22 blaise lol, I guess.. urhm,, 20:22 blaise sapier: I shouldn't be grabbing your animals_modpack from git.. eh? 20:23 blaise only the zip? 20:23 sapier you can grab it from git but you need adv_spawning too ... and factions is suggested too ... it's all written in lead post ;-) 20:23 blaise yeah, I know 20:23 blaise I got all that 20:23 blaise having random crashes though 20:23 sapier and you shouldn't grab the 2.5 dev version that's early wip 20:24 sapier can you provide the error logs blaise? 20:24 blaise can't seem to keep it up for any ammount of time 20:24 blaise yeah, just a sec 20:24 blaise should I wgetpaste for url, or drop in the chan? 20:25 blaise I'll wgetpaste 20:25 blaise lmao 20:31 RealBadAngel hi 20:31 RealBadAngel kahrl, youre here? 20:31 blaise Hi! 20:31 * blaise hugs RealBadAngel 20:31 RealBadAngel oh why? 20:34 blaise sorry.. 20:34 * blaise takes his hug back 20:34 RealBadAngel lol 20:36 Calinou hug takeback party 20:38 VanessaE_ sapier: *poke* 21:20 RealBadAngel quick question: why light tables start with 8? 21:20 RealBadAngel thx to that nodeboxes are lightsources 21:20 Calinou because full darkness doesn't make sense 21:20 Calinou yeah, if you use the debug light tables, anything that's not a solid cube will glow in the dark 21:21 Calinou this includes plantlike nodes and nodeboxes… 21:21 RealBadAngel blue channel is supposed to hold lightsources only 21:21 RealBadAngel light tables that start with 8 are fucked up 21:22 RealBadAngel Calinou, you would expect then stairs, slabs etc to glow in the dark? 21:22 Calinou nope 21:22 Calinou it's a bug, no idea why 21:23 RealBadAngel light table (in the current use at least) has to be revisited 21:23 RealBadAngel and 0 has to be 0 21:23 RealBadAngel no matter what 21:24 RealBadAngel in commented out ones, we can find such sets starting with 0 21:25 RealBadAngel by now, thx to that table, video::SColor c = MapBlock_LightColor(255, l, decode_light(f.light_source)); 21:26 RealBadAngel with f.light_source = 0 21:26 RealBadAngel blue channel holds 8 21:26 RealBadAngel and node becomes lightsource at level 8 21:27 RealBadAngel this is nodeboxes case 21:29 RealBadAngel Calinou, wasnt that you that wanted to change the light table? 21:30 VanessaE_ https://github.com/minetest/minetest/issues/1362 21:30 RealBadAngel indeed 21:32 Calinou yes