Minetest logo

IRC log for #minetest-dev, 2013-04-05

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:00 RealBadAngel i need to test the change first, i may have an idea how to do that
00:00 RealBadAngel i will play with it on weekend
00:01 VanessaE joined #minetest-dev
00:01 Taoki ok
00:01 RealBadAngel hi VanessaE
00:01 hmmmm yeah, it looks good
00:02 PilzAdam https://github.com/PilzAdam/minetest/commit/c5a8448c41e4ea9d33a43cebef61425d4568a46d
00:03 PilzAdam this works with nil -> not override current setting
00:03 Taoki Awesome. Looks good
00:03 hmmmm so you're handling that?
00:03 hmmmm thank you
00:05 hmmmm by the way pilzadam, when you commit something that has a pull request, be sure to paste the link to the commit and say something like "done" in a comment then close it
00:05 PilzAdam random idea: Lua side sprinting :-)
00:05 hmmmm uh no.
00:05 hmmmm horrible idea
00:05 PilzAdam :D
00:05 RealBadAngel PilzAdam, have  you read what i wrote bout creative tools?
00:05 VanessaE hey RB.
00:05 VanessaE A.
00:05 PilzAdam RealBadAngel, what exactly do you want?
00:05 hmmmm as for the "add support to latest C++ standard" pull request, i dunno, there are lots of compilers that don't support this
00:06 RealBadAngel imho you shall leave tools intact regardless the game mode
00:06 hmmmm i don't forsee us using any of those funky language features he was giving examples of
00:06 hmmmm it's a general rule (in minetest anyway) to make your code really stupid and simple
00:06 PilzAdam RealBadAngel, what you suggest is that the hand digging time shouldnt be used if the tool doesnt define it
00:06 Taoki PilzAdam: My change would allow that too :)
00:06 hmmmm and don't use any super-duper high level C++ constructs
00:06 Taoki The sprinting I mean
00:07 Taoki Anyway, need to go to bed since it's late. Waiting for the code to be ready and merged to be sure it's all ok
00:07 PilzAdam Ill add that to MiniTest :-)
00:07 Exio hmmmm: please, i need my object-matching-pattern-wtfisthatshit super-OOP things for live!!!!!!!111!
00:07 Exio hi btw :P
00:07 hmmmm and jesus christ.. C++11 allows you to overload quotation marks?
00:07 RealBadAngel PilzAdam, havent checked the code, just noticed fact that diamond swords digs concrete blocks faster than pickaxe
00:07 hmmmm okay it's beyond saving
00:08 PilzAdam RealBadAngel, the hand digs stone faster than a pickaxe, so does any tool that doesnt define the cracky group
00:08 hmmmm thanks to this guy's pull request, i know to specifically disallow C++11
00:08 Taoki MiniTest?
00:08 PilzAdam Ill push that now
00:08 Taoki ok
00:09 PilzAdam Taoki, yes, a game by me that copys MC
00:09 Exio hmmmm: lol
00:09 Taoki Fun fun
00:09 RealBadAngel PilzAdam, that should be somehow fixed then
00:09 PilzAdam the creative mode isnt meant to use tools
00:10 Exio what about the cosmetic fix for infinite tools?
00:10 Exio the stacks are already infinite, i don't see the "why not" of that
00:10 PilzAdam anything else that requires a protocol version bump?
00:10 Exio as it is just cosmetic
00:10 RealBadAngel folks sometimes want to play legacy on creative servers
00:10 RealBadAngel go on VanessaE server and see
00:11 Taoki PilzAdam: Thanks PilzAdam. Tomorrow I'll fix the ladder speed too since that's something else a lot of people complained about
00:11 Taoki Night
00:12 RealBadAngel night Taoki
00:12 PilzAdam hmmmm, I have this MC like sand falling (with a little delay when updating) and VanessaE said it is awesome
00:13 PilzAdam it doesnt add thousands of entities at once when digging a huge amount of flying sand
00:13 PilzAdam and it looks pretty nice
00:13 PilzAdam https://github.com/PilzAdam/minetest/commit/ea9fd59f2bce8d8a31f7bd4a952564faa0629441
00:13 VanessaE works on gravel too
00:13 VanessaE other falling nodes also, I'm sure.
00:14 hmmmm PilzAdam, sure, doesn't really matter to me though
00:14 hmmmm if you (actual minetest players) like it, then I guess add it
00:15 hmmmm just make sure others agree with it
00:15 hmmmm those are gameplay tweaks I can't really comment at all on
00:15 Exio https://github.com/sfan5/minetest/commit/30d029814ef5f0680510146d0e4b72bccd1bae50
00:15 Exio i would like to see this
00:16 PilzAdam hmmmm, it acutally isnt a gameplay tweak
00:16 PilzAdam Exio, it turned out that you cant apply more than one shader to a material IIRC
00:16 Exio ehm
00:17 PilzAdam http://irc.minetest.ru/minetest-dev/2013-04-04#i_2984467
00:19 Exio oh
00:20 Exio as doing the stuff, what do you think about https://github.com/minetest/minetest/pull/578 ? (it will make the "look" of the IRC-mod and others similar mods "better"
00:21 PilzAdam seems ok to me
00:22 hmmmm https://github.com/minetest/minetest/pull/596#issuecomment-15932423
00:23 Exio upgrade your compiler hmmmm
00:24 Exio the last gcc is super cooler(tm)
00:24 hmmmm in order for me to allow merging something, it'd have to *work* for me in the first place
00:24 hmmmm and I know it'd cause problems for other freebsd users
00:25 hmmmm freebsd is a tier-1 supported platform for minetest
00:25 PilzAdam pushed the mc like sand falling
00:25 hmmmm meaning it _has_ to work on there, or else there's a problem
00:26 PilzAdam hmmmm, I fixed the nick completion fix to not copy the list all the time: https://github.com/PilzAdam/minetest/commit/f245c6dc18e2b84d480209b64fec1fdfa8262a92
00:26 hmmmm i saw that
00:26 jordan4ibanez delay in node_update, won't that cause a lot of problems?
00:26 PilzAdam (yes, I fix fixes :-))
00:26 hmmmm why not push it?
00:26 jordan4ibanez why not a slow_node_update instead??
00:27 VanessaE PilzAdam: how short of a delay can you use and still have it work?
00:27 PilzAdam hmmmm, If you are ok with it now Ill push it
00:27 VanessaE (for the falling code)
00:27 hmmmm hmm
00:27 PilzAdam well, It had a delay of 0 and it worked, so this question isnt very good
00:28 VanessaE I ask because I'm concerned about those 0.1s delays adding up and causing lag is all.
00:28 PilzAdam I wouldnt call it "lag"
00:28 hmmmm it creates a nodetimer for each falling node, doesn't it
00:28 hmmmm that has potential to be bad if there's a large amount of nodes that start falling at once
00:28 PilzAdam hmmmm, no, it just queues them in globalstep
00:28 hmmmm ah
00:29 jordan4ibanez That's cool!
00:29 PilzAdam Ill push the nick completion now
00:29 jordan4ibanez left #minetest-dev
00:29 hmmmm oh shoot
00:29 VanessaE PilzAdam: re: nick completion, does that work in the 't' command line yet?
00:29 hmmmm there was something about that i wanted to make sure of before you push it
00:29 PilzAdam VanessaE, nope
00:29 VanessaE damn.
00:29 hmmmm you changed something from a wstring to a string
00:29 VanessaE that would be incredibly handy to have.
00:29 hmmmm are you sure that's correct?
00:30 PilzAdam player names are send to the client as std::string
00:31 PilzAdam so I keep it in the env as a string and change it to wsring when passing it to Irrlicht
00:31 hmmmm ah, it's irrlicht that expects the wstring
00:32 PilzAdam also, hmmmm, what do you think about the "Server -!-" prefix for chat_send_player()?
00:33 hmmmm i don't know
00:33 hmmmm that prefix was put there for a reason
00:33 Exio or at least making it less annoying for short messages or relays
00:33 hmmmm it's good to know what messages are coming from the server though
00:34 hmmmm maybe server messages could be drawn with a slightly different colored font
00:34 Exio -S- from the server? or something what is shorter than the word + signs
00:34 PilzAdam as a modder, I think its very annoying
00:34 PilzAdam Exio, nobody would understand that :-=
00:34 PilzAdam *:-)
00:34 hmmmm is that possible, though, the way the font works with irrlicht?
00:34 hmmmm if it is, i'd remove the prefix and change the coloring
00:34 PilzAdam wtf
00:35 PilzAdam github has a new GUI for pull requests
00:35 hmmmm yeah i saw
00:35 hmmmm it's what websites do
00:35 hmmmm they change things for the sake of changing things, not like it's being made better at all
00:35 VanessaE font?  push my bigger, sharper one ;)
00:35 hmmmm vanessae, not until the culling issues are fixed
00:35 PilzAdam VanessaE, we have TTF now
00:35 hmmmm also that
00:35 PilzAdam just install your own if you like to
00:36 VanessaE ttf sucks
00:36 VanessaE no shadowing.
00:36 VanessaE hmmmm: culling issues?  you mean kerning right?
00:36 PilzAdam you even have a font_size setting for ttf fonts ((c) PilzAdam)
00:36 hmmmm kerning yeah
00:36 VanessaE hmmmm: right.  I agree on that point.  Forgot about that issue actually.
00:36 hmmmm but i agree, the main purpose of the font change is to add that border for readaibility
00:37 hmmmm so increasing the font size of what's already there won't really help
00:37 hmmmm can we have a shadow?
00:37 hmmmm what if we write the same string twice, one in black, offset by 1+1 pixel?
00:37 hmmmm i'd actually rather not do that, i'm just throwing stupid ideas out there
00:38 VanessaE hmmmm: you'd need to actually outline the whole font
00:38 VanessaE a 1x1 offset only shadows part of it, wouldn't work in practive
00:38 VanessaE practice* .  I tried already.
00:39 PilzAdam oh fuck, its too late again :-(
00:42 VanessaE ?
00:42 VanessaE oh
00:42 VanessaE <PilzAdam> bye
00:42 PilzAdam how about that one? https://github.com/PilzAdam/minetest/commit/2126bf06d20f6cb5bb6a0dbe728923b228d5c5cb
00:42 VanessaE (kidding)
00:42 VanessaE mm
00:42 VanessaE you need to widen the list box for that
00:43 VanessaE well, maybe not actually.
00:44 VanessaE in my particular use case, doing that would cause about half the entries in my worlds list to run out past the right margin
00:44 PilzAdam even minimal with the long name fits into it
00:44 PilzAdam and it is hardly at the half
00:44 VanessaE probably not a real concern then
00:44 PilzAdam (with the worldname being "minimal")
00:46 PilzAdam well, its a bit more critical in the advanced tab
00:47 PilzAdam but using the foldername instead of the actual gamename is stupid IMO
00:47 dexter0 joined #minetest-dev
00:48 PilzAdam hmmmm, what do you think?
00:59 hmmmm i think we shouldn't take up unnecessary space
01:00 hmmmm i just hope it isn't too wide
01:02 salamanderrake joined #minetest-dev
01:11 mrtux_ joined #minetest-dev
01:28 mrtux joined #minetest-dev
01:28 mrtux joined #minetest-dev
01:33 hmmmm hah wow
01:33 hmmmm https://github.com/Bukkit/mc-dev/blob/master/net/minecraft/server/BiomeBase.java
01:33 hmmmm coincidentially, i limited the number of biomes to 256 too
01:33 Exio i will not open *that* file
01:34 hmmmm except the first one is occupied by a "default" biome that exists no matter what
01:34 VanessaE I can't see us needing more than a dozen biomes anyway
01:34 hmmmm and it seems that they have something that's similar to DecorationDef, and they call it BiomeDecorator
01:34 hmmmm i want to check that out, i want to know if it really is what i think it is, and if they do the placement like i do too
01:35 hmmmm nevertheless, i'm going to try to rip their biome temperature/humidity settings so we have something that doesn't need balancing
01:36 hmmmm it might be tougher than that, because i don't do the max/min like they do
01:38 hmmmm also i think the mushroom island idea is stupid and i'll omit that biome altogether
01:38 hmmmm https://github.com/Bukkit/mc-dev/blob/master/net/minecraft/server/BiomeHell.java   gotta really love Java with their one-class-per-file restriction
01:39 Exio oh god, stop posting links what finish with .java
01:39 hmmmm you'll live
01:39 Exio if i open those links i'm sure i'll start bleeding
01:48 hmmmm sand, gravel, ore, flowers, mushrooms, huge mushrooms, reed, cactus, waterlilly
01:49 hmmmm they handle trees separately for some reason, it seems
01:50 VanessaE oh G*d no, not those huge mushrooms
01:50 VanessaE what idiot thought those made even a lick of sense?
01:51 VanessaE oh wait, Notch did. ;)
01:51 Exio the only thing what i miss of MT actually is the nether (as the only think i do in MC actually, is playing in the nether)
01:51 hmmmm or probably one of the designers he employed
01:52 Exio i don't like them for the mushrooms, but mobs can't spawn in those biomes
01:52 ShadowNinja I like how descriptive calling a() is
01:52 Exio (on those?)
01:52 hmmmm it's not hard to do that
01:52 Exio hmmmm: yep, just saying "what i see is good for those biomes" :P
01:53 ShadowNinja you can use multiple characters for variables!
01:53 hmmmm when it comes time for mobs, we'll just leave some biomes without them
01:53 hmmmm shadowninja, that's not really the source, that's just decompiled
01:53 Exio the code of bukkit is "de-ofuscated" code
01:53 Exio er, that
01:54 Exio the *real* code is not available for public
01:54 ShadowNinja hmm, ok fine
01:56 hmmmm the way they do beaches is kind of smart
01:56 hmmmm https://github.com/Bukkit/mc-dev/blob/master/net/minecraft/server/WorldGenSand.java
01:57 hmmmm pick a random radius, at least 2, replace all the blocks around that are water to be sand
01:57 hmmmm doesn't require any perlin noise that way
01:59 hmmmm and the way they do nether is different from what i expected, and i'd have to say more difficult.. i'm pretty sure negative 3d perlin noise space would've worked just as well
02:03 hmmmm oh, ahh, those pieces are for the nether fortress, not the nether itself
02:03 hmmmm that's an interesting approach to procedurally generating large structures, but i'd have to say ultimately boring.  L-systems beat this
03:20 kaeza1 joined #minetest-dev
04:12 VanessaE hrm.  the "build" and "survival" games are not included in the 'make install' step?
04:37 kaeza aren't they installed separately just as minetest_game?
04:38 VanessaE well they can be, but minetest_game is covered when you `make install`
04:41 VanessaE I kinda figured the other two would be also.
04:49 kaeza then that is inconsistent at least
04:49 VanessaE yes.
04:50 VanessaE I'd like my builds to include those two games, but it's not possible without help from one of the upstream devs.  I don't understand shit when it comes to makefiles :)
04:53 kaeza1 joined #minetest-dev
04:57 kaeza1 https://github.com/minetest/minetest/pull/600
04:58 VanessaE \o/
04:58 kaeza hope that helps
04:59 kaeza ;)
04:59 VanessaE cherry-picked.
04:59 kaeza fux
04:59 kaeza forget about that
04:59 VanessaE let's see what THIS breaks
04:59 VanessaE shit.
04:59 VanessaE ok
04:59 kaeza I meant to edit my repo, and edited minetest master instead
05:00 kaeza sorry about that
05:00 VanessaE reverted.
05:00 VanessaE eh?  it looks to me like it's on your repo?
05:03 kaeza ah sorry lol I meant the pull request
05:03 kaeza https://github.com/kaeza/minetest/commit/2de3956cb0787231d87e79cff0178d60af8ac764
05:04 kaeza can you test that?
05:04 VanessaE sure, sec
05:05 kaeza I forgot to change line 153 to .../survival in the previous one
05:05 VanessaE ah
05:05 VanessaE running it through my regular build script, let's see what this breaks ;)
05:05 * kaeza crosses fingers
05:07 VanessaE hrm, something didn't work.
05:07 VanessaE lemme check my script.
05:09 kaeza what's the error?
05:09 VanessaE my script didn't install minetest_game et.al before cmake was executed.
05:09 kaeza ah
05:09 VanessaE re-running it now with that change.
05:10 kaeza I'll need to rework that change to use a foreach loop
05:10 kaeza may be useful in case someone decides to add new "standard" games in the future
05:10 VanessaE eh, leave it unrolled
05:11 VanessaE make install coming up in seconds... *crosses fingers*
05:11 VanessaE \o/
05:11 kaeza success?
05:11 VanessaE I see all four components in the make install spew
05:11 kaeza \o/
05:11 kaeza make pull request?
05:11 VanessaE hold, let me verify it
05:14 VanessaE I think it's good
05:14 VanessaE uploading the build to my web space now.
05:14 kaeza ok
05:14 VanessaE http://forum.minetest.net/viewtopic.php?pid=81578#p81578
05:15 VanessaE can you double-check the output?
05:15 kaeza sure
05:22 VanessaE *waits*
05:23 kaeza still building...
05:23 kaeza I usually compile with RUN_IN_PLACE
05:23 VanessaE ok
05:28 kaeza success!
05:29 VanessaE \o/
05:29 VanessaE pull-request that sucker. :)
05:35 ecube joined #minetest-dev
05:40 kaeza I need some help with this
05:41 VanessaE I always do that from the website
05:41 kaeza I accidentally pushed to my master branch, but I already have a pull rq for it
05:41 VanessaE that's ok, it'll get refreshed to the new code.
05:41 kaeza no I mean, the pull is for another unrelated thing
05:41 VanessaE oh
05:42 VanessaE then you'll have to push the change to a different branch and pull-request that against upstream master
05:42 VanessaE https://github.com/minetest/minetest/pull/593
05:42 kaeza I did that, but now all the changes (including the commit in the open pull) are in it
05:42 VanessaE there it is.
05:43 VanessaE in that case you have to revert your commit and push -f
05:43 kaeza you know what? I'll nuke and reclone
05:43 VanessaE or just clo.......
05:43 VanessaE or do that.
05:43 VanessaE :D
05:43 kaeza :P
05:54 kaeza https://github.com/minetest/minetest/pull/601
05:54 kaeza :D
05:54 * kaeza is starting to love rm -fr
05:54 VanessaE heh
05:57 celeron55 Taoki|away: why did you add the implementation and declaration of gob_cmd_update_physics_override to that position in the file while all the others were in the same order as in the enum? https://github.com/minetest/minetest/commit/c5a8448c41e4ea9d33a43cebef61425d4568a46d
05:58 celeron55 (actually not enum but bunch of #defines but anyway)
06:02 nyuszika7h_ joined #minetest-dev
06:03 VanessaE kaeza: better. :)
06:04 kaeza hah. that's the powa of git push -f!
06:04 VanessaE hehe
06:05 kaeza and now back to my side-project of bringing Puppy builds up to date
06:06 VanessaE cool
06:06 kaeza latest in the repos is 0.4.3 :/
06:07 VanessaE better than the ubuntu repos, which still hold 0.3.x
06:52 ImQ009 joined #minetest-dev
06:54 BackupCoder joined #minetest-dev
08:53 Calinou joined #minetest-dev
08:57 Calinou joined #minetest-dev
09:05 serengeor joined #minetest-dev
09:27 ImQ009 joined #minetest-dev
09:38 Calinou joined #minetest-dev
09:40 Calinou_ joined #minetest-dev
09:52 troller joined #minetest-dev
10:16 Taoki joined #minetest-dev
10:34 Taoki celeron55: Thought I added the new function exactly like all others. At least that was my intention
10:37 Taoki Aaah... the #define stuff. Yeah... I added that one at the bottom cuz you said not to invert and modify numbers any more. Else I could have centered it as well
10:38 Taoki I tend to prefer adding things in the order that makes sense, not necessarily "newest at the bottom". I'm too perfectionist when coding xD
10:39 Taoki You can re-arrange it in the enum too if it's safe, but I think that's where you said never to modify old numbers cuz it breaks previous clients. Yeah... I should have just added everything new to the bottom really
10:43 PilzAdam joined #minetest-dev
10:54 proller joined #minetest-dev
10:57 kaeza joined #minetest-dev
11:09 jojoa1997|Tablet joined #minetest-dev
11:20 jojoa1997_Tablet joined #minetest-dev
11:22 jojoa1997|Tablet joined #minetest-dev
11:29 jojoa1997_Tablet joined #minetest-dev
11:29 ImQ009 joined #minetest-dev
11:43 jojoa1997_Tablet left #minetest-dev
11:46 Calinou joined #minetest-dev
11:47 Jordach joined #minetest-dev
11:47 Jordach joined #minetest-dev
12:40 rarkenin_ joined #minetest-dev
12:48 Taoki celeron55, PilzAdam: I fixed the aux1_descends problem that everyone was complaining about. https://github.com/minetest/minetest/pull/603
12:48 Taoki The fast movement on ladders problem
12:49 Taoki VanessaE: ^
12:49 PilzAdam localplayer.cpp:192 this line is way too long
12:50 Taoki Never heard about lines which are too long before :P Relocate the comment at the end in that case
12:50 PilzAdam lines shouldnt contain more than 80 chars and ~100 is the maximum
12:51 Taoki That comment is important so people always know exactly why it's done that way. But move it if needed
12:53 Taoki Anyway I tested locally both with and without aux1_descends and it works as intended
12:57 PilzAdam yes, it works
12:59 PilzAdam also this commit message is too long
13:00 Taoki Aaaargh :P
13:00 Taoki Even that is harmful
13:00 PilzAdam pushed
13:01 Taoki thanks. For the future I'll try to keep stuff shorter if it's important
13:21 hmmmm joined #minetest-dev
13:24 iqualfragile joined #minetest-dev
13:38 serengeor joined #minetest-dev
13:39 Taoki I wonder if anyone else things we should have mapgen floating islands by default in minetest_game...
13:39 Taoki **thinks
13:39 Taoki Tried out the floatlands mod and it's beautiful. Really wouldn't harm to have it default.
13:39 Taoki Once it's no longer so layyg and slow to generate that is
13:40 PilzAdam use mg_name = indev
13:40 PilzAdam it has floating islands
13:46 PilzAdam anything against this? https://github.com/minetest/minetest/pull/553
13:48 RealBadAngel nothing
13:49 Jordach no problems in my eyes
13:50 Taoki nope, looks good
13:51 Taoki And what's mg_name - indev? Where and what is that?
13:51 PilzAdam its a setting in minetest.conf
13:51 RealBadAngel Taoki it shall be "nAsWs/c&aux1_d is on"
13:52 RealBadAngel and everybody will know from start what you meant ;)
13:52 Taoki oh, ok. What does it do exactly?
13:54 Taoki What is mg_name = indev exactly though? It enables experimental features in the code?
13:55 PilzAdam we have support for mutlitple map generators in the engine
13:55 PilzAdam the default is mg_v6
13:55 PilzAdam indev is the playground of proller
13:56 Taoki oh nice
13:57 Taoki Still unclear what the statuus is on defining biomes in LUA btw. And getting rid of things like alias_dirt / alias_sand / etc. Of course while the mapgen stays in C++
13:58 PilzAdam hmmmm is working on magpen v7
13:58 PilzAdam it will have all this register_biome() stuff
13:58 Taoki Awesome!
14:00 Taoki Hoping it would be something like "define areas where these types of nodes are used, and have this amount of caves, tunnels, rivers, etc"
14:01 Taoki Also, I find the indev mapgen a lot prettier :)
14:02 Taoki It generates some amazing places. Default mapgen does too but this one even better
14:02 Taoki However, before I see floating islands I see a few floating trees. And no, not in the good sense :P
14:02 Taoki (very rarely still)
14:03 PilzAdam the only difference is that it has larger mountains at the edge and floating islands
14:04 proller and CAVES
14:05 proller and biomes larger
14:06 Taoki I'm quite above cloud height and see no islands though
14:06 PilzAdam you have to go to 500+
14:06 proller 500+ - very rare
14:06 proller 30000 - more islands
14:07 proller go to "sky hardcore" server -
14:07 Taoki Wow, that's way high >.<
14:07 proller spawn at 30000
14:07 Taoki How do you do that?
14:07 proller what ?
14:08 proller static_spawnpoint = 15,29539,-8
14:08 proller mg_name = indev
14:08 rarkenin joined #minetest-dev
14:08 PilzAdam what mods do you have there?
14:08 Taoki Also, MineTest seems to be bacly lagging the system when in view. Can barely even alt-tab switch although it's windowed
14:08 proller bones elevator minetest-teleporters moretrees nuke plants_lib sponge tnt
14:09 Taoki proller: I meant how to teleport at that height
14:09 proller /teleport 0,30000,0
14:10 Taoki thanks, worked
14:10 Taoki Hmm. Could mods make floating islands be made out of other materials than classic dirt and stone? Also, can you change the spawn height of islands?
14:12 Jordach joined #minetest-dev
14:12 Jordach joined #minetest-dev
14:12 proller # Float islands starts from height, 0 to disable
14:12 proller #mgindev_float_islands = 500
14:13 proller about adding new materials - ask PilzAdam
14:14 Taoki ok
14:15 proller i think about some kinds of float lands, maybe something like walkable clouds
14:15 proller and like asteroids from stone or metal
14:38 Taoki Trees with black leafs are a nice species of tree that got implemented in MineTest ;)
14:38 Guest27058 joined #minetest-dev
14:39 kaeza joined #minetest-dev
14:40 kaeza I see you are not able to build the executables into a directory different than the source dir; minetest and minetestserver are always output to minetest/bin
14:40 kaeza is there any particular reason for this?
14:41 kaeza I see you are not able to build the executables into a directory different than the source dir; minetest and minetestserver are always output to minetest/bin
14:41 kaeza derp
14:41 kaeza I see you are not able to build the executables into a directory different than the source dir; minetest and minetestserver are always output to minetest/bin
14:41 kaeza I'm editing the CMakeLists.txt file to be able to do it, but asking just in case
14:50 Calinou joined #minetest-dev
14:50 Jordach afternoon Calinou
14:53 Zeg9 joined #minetest-dev
15:00 rubenwardy joined #minetest-dev
15:06 proller about floatlands - now we cant build stairs down
15:06 rubenwardy left #minetest-dev
15:08 proller you cant buld anything below lower block of float land
15:11 kaeza can someone test this: https://github.com/kaeza/minetest/commit/70a23c2a7898df310b651cc4cf1a50bddb7ba2d6
15:11 proller http://paste.org.ru/?9sk33e
15:11 kaeza I tested and works as expected
15:15 PilzAdam proller, you can, with water and lava
15:15 proller where get water or lava in sky?
15:16 PilzAdam doesnt it spawn there?
15:17 proller i tested some water springs at sky, but now waiting hmmmm while he fix ore gen with param1 param2
15:18 proller and springs make now very good lakes in air above unloaded blocks
15:18 proller not good
15:20 proller possible fix - delete liquid when it falls on unloaded node
15:20 proller above sea level only
15:20 PilzAdam there was a fix for it IIRC
15:25 Calinou proller: suggestion: add nodes that do 20 damage per second below floating island
15:25 Calinou 16 layers of them or so :P
15:27 celeron55 Taoki: you didn't understand at all; it's not about what the code does; it's a bout how it's organized
15:27 celeron55 -" "
15:27 Jordach about*
15:30 proller Calinou, for what? you die if fall anyway, but now is bug - you can stop while falling, and land on next island with low speed
15:32 Calinou stop thinking like a debian user; it takes way too long to die
15:32 Calinou or bundle a /kill command with your mod :P
15:36 celeron55 the the fuck kind of new spam is this http://paste.dy.fi/9zr
15:36 Calinou lol :D
15:43 Jordach celeron55, huh, changing from viagra (with virus) to now mobile phone games which read your address book and spam the fuck out of you and your friends
15:56 celeron55 i consider anyone who install such a game very stupid
15:57 celeron55 also, i've always wondered how the viagra crap can possibly spread any computer viruses
15:57 celeron55 like... do people, like, think at all?
15:58 Jordach celeron55, not anymore, this is the age of instant gratificartion
15:58 Jordach -r
15:58 Jordach where people are essentially sheep
16:02 celeron55 i've always wondered if i should make a re-branded "minetest for sheep"
16:02 celeron55 it could be a hit
16:02 celeron55 :D
16:04 Jordach celeron55, go ahead
16:05 dexter0 joined #minetest-dev
16:09 celeron55 it'd need to be a very stripped down and polished version with some very simple focus and a installer that is closer a virus than an actual legit installer
16:11 celeron55 i'm supportive of this as an experiment; but not really ideologically
16:11 Jordach see how dumb people are by giving a zipfile which just contains a link to a survey download
16:12 Jordach (some download sites have the downloader to answern surveys which never answer)
16:13 kaeza point them to this: http://nooooooooooooooo.com/
16:13 * Jordach claps
16:53 rubenwardy joined #minetest-dev
17:10 Taoki celeron55: Yeah. I meant I like to keep it clean and organize everything in order, but I should have added this at the bottom instead
17:17 celeron55 yes, that is what i mean; while the code does what it should, it's not in perfect order while it could as well be
17:18 PilzAdam why isnt launchpad updated to 0.4.6 yet?
17:18 celeron55 because i haven't remembered
17:19 celeron55 probably somebody other should register there and let me teach how to use it
17:20 celeron55 also, launchpad isn't updated to have tracking bazaar repos for the other games
17:20 celeron55 it needs to be done via support tickets
17:20 rubenwardy1 joined #minetest-dev
17:21 celeron55 launchpad is a fucking pain overally, i don't really understand why i try to tolerate it
17:21 celeron55 anyway, i'll make it make a 0.4.6 with minetest_game
17:22 Taoki celeron55: Anyway, I fxied the annoying ladder speed problem today. Sorry it took so long, but now everything's fine at last for aux1_descends users :)
17:24 celeron55 Taoki: i saw that
17:24 celeron55 haven't tried yet though
17:25 Taoki Thinking of other things for the engine and which of them I can help with doing
17:26 celeron55 actually... now that i think of it, i think you *can* make tracking repos in bazaar via an online form; it's the changing of them that needs support tickets
17:26 Taoki BTW, tried the floating islands for the first time (the C++ ones in the dev mapgen). Really lovely stuff, can't wait for them to be defaulted on :)
17:26 celeron55 but eh, no, i'm not going to do that
17:27 celeron55 why can't launchpad support plain git repos and get stuff based on tags in them; it would be like million times easier
17:30 Taoki Who handles the mapgen floatislands BTW? I have a suggestion if they're around
17:30 Taoki (haven't seen a forum topic for these, just for the LUA floatlands mod)
17:32 celeron55 i think proller is the one messing around with the "indev" mapgen
17:32 rubenwardy joined #minetest-dev
17:33 celeron55 yeah, the forum isn't really used for core stuff development, except by some people
17:33 proller its port of floatlands mod with minor changes
17:33 celeron55 this channel and github are better for tracking such
17:34 proller but dislike this algorithm, it uses 3 differend perlins, but islandg looks good
17:35 proller ups, misses..
17:36 Taoki oh, ok :)
17:36 PilzAdam celeron55, I have set up an launchpad account
17:37 Taoki proller: My question was if you could add water to float islands too. In some cases near the edge, so that waterfalls will go down and even reach the ground eventually (might not happen due to the huge distance but still)
17:38 proller ^^^ <proller> and springs make not very good lakes in air above unloaded blocks
17:38 proller ^^^ <proller> i tested some water springs at sky, but now waiting hmmmm while he fix ore gen with param1 param2
17:39 proller springs was broken in new oregen
17:39 proller but possible add with old lua function
17:39 rubenwardy1 joined #minetest-dev
17:39 rubenwardy1 left #minetest-dev
17:41 Taoki proller: Another suggestion; The version paramat made uses special nodes for floating islands (float-stone and float-dirt which is dirt of a more cyan color). Do you think the mapgen version should do the same? Would be nicer than using the same type of stone and dirt as the normal ground
17:42 proller and i think visual of flowed water must be remaked, its too buggy now
17:42 rubenwardy joined #minetest-dev
17:43 proller i think i must go home right now, will answer later
17:43 celeron55 PilzAdam: if you have any questions, i recommend asking before doing
17:43 Taoki ok
17:44 celeron55 PilzAdam: also, ask too if you don't have anything to ask, because then i am guessing you just don't know what to ask (because nobody can know how to use launchpad before having done or heard all the mistakes once)
17:44 Taoki BTW. Was there some debug key to enable wireframe mode? I forgot, and wanted to check something
17:51 celeron55 Taoki: no
17:51 celeron55 (afaik)
17:51 Taoki ok
17:52 Taoki celeron55: I wanted to check if block and nodebox surfaces are drawn as triangles or quads, for performance. Not sure if Irrlicht allows this itself, but IIRC if the surface doesn't generate any spikes, it can be drawn out of quads instead of triangles which might be faster.
17:52 Taoki Since the surfaces of blxes / voxels are flat this would work
17:53 Taoki IIRC pure OpenGL allows it. You can define a quad surface
17:53 PilzAdam celeron55, the code gets imported automatically into the bazaar repo?
18:11 celeron55 PilzAdam: yes
18:12 celeron55 then dailies automatically built from it; the stable packages are built according to the recipe (which you can probably find), that needs to be updated for each version and then the new build must be requested
18:12 celeron55 to make a new stable build, three things must be changed: the package version, the minetest bzr revision and the minetest_game bzr revision
18:13 celeron55 the ppaX at the end of the package version is for new launchpad builds of the same minetest version (because you screw them up half of the time)
18:14 celeron55 all builds use the minetest_packaging or something like that repo that contains all the debian packaging things, including the command line for building and the package description
18:15 celeron55 generally the daily builds don't need to be tuned at all; i guess that needs to be done only when the packaging repo is updated, or if more games or something would be added
18:18 celeron55 complicated enough? 8)
18:19 PilzAdam updating the recipe for stable just means setting the revno to the commit?
18:21 PilzAdam and how would one set up the bazaar branches for common, survival and build?
18:22 celeron55 somewhere there is some "import external repository" thing
18:22 PilzAdam hmm... the stable build you made shouldnt work because you are missing common
18:22 celeron55 that's very true... didn't even remember 8D
18:22 celeron55 as you can see, that was ppa0
18:22 PilzAdam same for unstable
18:23 PilzAdam *daily
18:23 celeron55 beware that if you screw up the external import, you need to contact launchpad support to modify it
18:24 celeron55 i feel bad for anyone trying to package minetest 0.4
18:25 celeron55 0.3 is like the easiest possible, 0.4 is like the hardest possible
18:25 * PilzAdam cant find the import button
18:25 PilzAdam this interface sucks
18:26 celeron55 hmm
18:27 celeron55 i'm sure i found it once
18:27 proller joined #minetest-dev
18:28 celeron55 apparently it isn't automated: https://help.launchpad.net/Code/Imports
18:29 celeron55 this is the request form https://code.launchpad.net/+code-imports/+new
18:29 celeron55 i think we may need to implement gamepacks 8)
18:29 celeron55 it's too complicated to manage multiple game repos in some situations
18:29 celeron55 like this
18:30 PilzAdam should I fill these forms for common, build and survival?
18:30 celeron55 i guess so
18:31 PilzAdam what is a "branch" in bazaar?
18:31 PilzAdam is it master or common?
18:32 rarkenin joined #minetest-dev
18:32 celeron55 umm
18:32 celeron55 you mean the "Branch Name:" thing in that form?
18:32 PilzAdam I would guess its "common"
18:32 PilzAdam yes
18:32 celeron55 leave it as is
18:33 celeron55 that is "trunk"
18:33 celeron55 i don't understand what that does but at least "common" or "master" makes zero sense
18:33 PilzAdam well, here is the list of branches: https://code.launchpad.net/~minetestdevs
18:34 PilzAdam and they are named "minetest-c55" etc.
18:34 celeron55 ah
18:34 celeron55 wtf
18:34 PilzAdam and it says that it will be used in the URL to identify the branch
18:35 celeron55 name them after the github repos then
18:35 celeron55 and set the project to "minetest-c55"
18:36 celeron55 and set the owner to minetestdevs
18:37 PilzAdam https://code.launchpad.net/~minetestdevs/minetest-c55/common
18:39 PilzAdam ummm
18:39 PilzAdam the whole layout has changed: https://launchpad.net/minetest-c55
18:39 celeron55 i have a feeling this is going to explode somehow
18:43 PilzAdam oh, seems that was the site for the 0.3.1 packages that are in the official ubuntu repos
18:45 PilzAdam the minetestdevs site still looks the same and has the common branch in it (unimported yet)
18:45 PilzAdam I make the forms for the other games now
18:49 VanessaE PilzAdam: where was it that you added that little delay that made falling sand/gravel work?
18:49 PilzAdam in builtin/falling.lua in the nodeupdate() function
18:50 VanessaE thought so.
18:50 VanessaE regression warning:  this causes a massive lag when using worldedit //fixlight
18:51 VanessaE I suggest not adding it there, but rather to the part of the code that causes each sand/gravel block to start falling, if that's possible.
18:52 VanessaE (e.g. to some place where it won't be executed by mods that force nodeupdates for other reasons)
18:52 PilzAdam why?
18:53 VanessaE because the point of adding that delay was so that gravel/sand falls properly, but it horribly kills performance for things that need nodeupdate but don't care about the falling code.
18:54 PilzAdam umm... what?
18:54 PilzAdam nodeupate() only handles falling nodes
18:54 VanessaE 81000 nodes updated with my last //fixlight command, which uses dig_node() on air, which forces a nodeupdate.
18:54 VanessaE which means 8100 seconds for the "rest" of minetest to pick up after worldedit finishes.
18:55 VanessaE this used to only take 10 seconds or less
18:55 Jordach joined #minetest-dev
18:56 VanessaE I'll revert that commit from my local copy and rebuild to be sure.
18:57 PilzAdam well, nodeupate_single() is called for the same amount of nodes as before
18:58 PilzAdam the only difference is that only 1 of the 9 nodeupdate_single() calls is done instantly
18:58 PilzAdam the other 8 are queued to on_globalstep()
18:59 VanessaE same amount of nodes, but now more time per node
18:59 VanessaE with a few it's fine, with large amounts the time stacks up exponentially
18:59 PilzAdam so the first 8100 nodeupdate_single() calls are done instantly, and the other 64800 are done 0.1 seconds after that
19:00 VanessaE yeah
19:00 VanessaE that's 6480 seconds.
19:00 PilzAdam before that all 72900 calls were done instantly
19:00 VanessaE that's a bit less than two hours
19:00 PilzAdam <VanessaE> that's 6480 seconds. <- wrong
19:00 VanessaE before, all 81000 calls were done in maybe 10 seconds or less.
19:00 PilzAdam they are all done in the same environment step
19:01 VanessaE um
19:01 VanessaE no
19:01 PilzAdam the delay isnt added for every single node
19:01 VanessaE ok, I just rebuilt without your commit
19:01 VanessaE let's see how it handles now.
19:01 PilzAdam you dont have to rebuild
19:02 PilzAdam builtin can be modified without rebuilding
19:03 rarkenin joined #minetest-dev
19:03 VanessaE ok I just proved it
19:03 VanessaE without your patch, it takes 6 seconds to update 85000 nodes.
19:03 VanessaE with your patch, it goes on endlessly.
19:03 PilzAdam it is simply no difference in do the whole work in at once, or doing 1/9 and the other 8/9 after 0.1 seconds
19:03 VanessaE I just tested it.
19:03 VanessaE your code causes a massive regression.
19:04 PilzAdam this cant be
19:04 VanessaE I just timed it.
19:05 VanessaE I reverted your patch, recompiled (yeah I forgot I didn't need to), restarted, and executed a //fixlight on a slightly larger volume (closest I could get).
19:05 VanessaE it completed in 6 seconds.
19:05 PilzAdam are you sure that it doesnt stop with the commit?
19:05 VanessaE 100% positive.
19:05 PilzAdam that doesnt make sense at all
19:06 VanessaE maybe not, but that's where it stands
19:06 PilzAdam do you have LuaJIT running?
19:06 VanessaE maybe you're just not understanding when/how that function is being called or something
19:06 VanessaE yes I do, of course.
19:06 proller joined #minetest-dev
19:08 PilzAdam can you try it with a delay of 0 in minetest.after()?
19:09 VanessaE sure
19:10 VanessaE wait, where is that at?
19:10 PilzAdam falling.lua:179
19:11 VanessaE oh, that's with your patch in place heh.
19:11 VanessaE I already reverted it.
19:11 VanessaE just try it yourself, it's really easy to spot.
19:11 VanessaE take out your patch.  Run worldedit, select at least a 50,000 node region, //fixlight
19:12 VanessaE then put your patch in and do exactly the same thing.
19:12 PilzAdam celeron55, anything else todo than adding "nest common lp:~minetestdevs/minetest-c55/common games/common" to the recipes?
19:12 PilzAdam (same for the other games)
19:12 PilzAdam VanessaE, I dont have worldedit
19:12 VanessaE PilzAdam: so install it?
19:12 PilzAdam Im fighting with launchpad currently
19:13 VanessaE ok
19:13 VanessaE I recommend you revert that commit from upstream.
19:21 celeron55 PilzAdam: dunno, try 8)
19:23 PilzAdam should I also request a daily build?
19:24 VanessaE PilzAdam: in addition to default fire and leafdecay, a lot of external mods directly call nodeupdate also...though I don't see that being done in worldedit.
19:24 VanessaE somehow the dig_node() command must be calling it.
19:24 PilzAdam yes, its called in minetest.register_on_dignode()
19:24 VanessaE that's why.
19:24 Calinou joined #minetest-dev
19:25 VanessaE each call of minetest.after() causes a delay of x seconds
19:25 PilzAdam nope
19:25 VanessaE and if you //fixlight, you're executing dig_node() for every air node therein.
19:25 PilzAdam that is wrong
19:25 celeron55 PilzAdam: you can request a daily build anytime; they're quite disposable
19:26 celeron55 they aren't archived anywhere or anything
19:26 VanessaE wrong or not, directly or indirectly invoking nodeupdate() is the only possible way to force a lighting update.
19:26 VanessaE you need to add that delay to the part of the falling code that calls nodeupdate() in the first place.
19:27 VanessaE not to nodeupdate() itself
19:27 celeron55 nodeupdate() is not really in the API anyway
19:27 PilzAdam celeron55, it is
19:27 PilzAdam since its moved to builtin
19:28 celeron55 ehm... well i guess so then
19:30 VanessaE well, whether it's in the API or not, you busted it. ;)
19:30 celeron55 but the fact that worldedit even calls it is very wrong
19:30 celeron55 it doesn't need to be supported
19:32 VanessaE worldedit uses dig_node.
19:32 celeron55 dig_node isn't meant to work for such bulk operations
19:32 VanessaE anything that calls dig_node() is affected.
19:33 VanessaE well how else do you expect a mod to be able to force a lighting update to fix bad shadows?
19:33 celeron55 set_node is the only one supported when you have to do >5 node changes in one go
19:34 VanessaE there is no method I know of in the API (or not) that can cause a node's lighting to be updated except for dig_node() and the equivalent placement functions.
19:34 celeron55 VanessaE: fix lighting bugs in the first place, or add an api function for recalculating lighting in some place
19:34 celeron55 it isn't rocket science and both of these are obvious
19:34 VanessaE but that's not an option here.
19:35 celeron55 digging nodes when you want to recalculate lighting is nonsense
19:35 VanessaE this already worked great before pilzadam's patch
19:35 VanessaE his patch broke it, and that's the end of it.
19:35 celeron55 your way of always preserving compatibility over making things right is nonsense, ridiculous and will lead this project doom
19:35 celeron55 i recommend stopping it, or at least understanding that nobody should listen it
19:35 celeron55 +to
19:36 proller joined #minetest-dev
19:36 VanessaE no, what will lead to the death of this project is constant breaking of things that worked perfectly before.
19:36 celeron55 no it won't
19:36 celeron55 not in the scale we're doing it
19:37 celeron55 modders should more often implement useful features to the engine instead of working around them with hacks
19:37 VanessaE jesus H.  I'm talking about a, what was it, 4 line change to nodeupdate() that breaks a dozen mods for no benefit at all.
19:37 celeron55 that way these can be avoided altogether
19:38 celeron55 i'm pretty sure there is some reason for such a change
19:38 VanessaE which makes more sense?  rewrite a dozen mods, or move a couple of lines in nodeupdate() ?
19:38 celeron55 depends on reasons
19:39 VanessaE celeron55: I'm talking about a several minutes' delay in executing a function that used to take a few seconds.
19:39 VanessaE THAT is reason enough not to allow that change.
19:39 VanessaE the delay was added to fix a bug in falling code.  Put the change where it will only affect the falling code, not where it'll affect a bunch of other shit
19:40 proller Taoki, i think its bad to add random new blocks, we need something like Story with relations between blocks
19:40 VanessaE then if there's some one random mod that needs that delay, that one person can add it him/herself.
19:40 Taoki proller: Not random new blocks. Just a special type of smooth stone and dirt to use for floating islands
19:40 Taoki I like the cyan sand the LUA version uses
19:41 VanessaE (which, at present, is only the falling code, to my knowledge, though mesecons has minor timing glitches also)
19:41 Taoki Called float-sand and float-stone
19:41 celeron55 VanessaE: i guess you need to find a proper solution with PilzAdam then
19:41 VanessaE I just found one.
19:41 proller Taoki, for example like avatar idea about float lands, simple stone with rare ore with special properties (flying)
19:42 VanessaE dig_node() worked perfectly fine before the way it is being used
19:42 VanessaE there was no reason to bust it like that
19:42 Taoki proller: Nah, simple stone is kinda dead. It's nice to have smooth stone and dirt. But a slightly different type
19:42 celeron55 updating lighting with dig_node() makes me puke
19:43 VanessaE celeron55: well, it works.
19:43 PilzAdam VanessaE, I tested it now
19:43 VanessaE and it works perfectly fine.
19:43 celeron55 why won't remove_node work?
19:43 celeron55 or whatever it is
19:43 VanessaE celeron55: because fixlight needs to operate on AIR.
19:43 PilzAdam //fixlight takes exactly the same time as before
19:43 proller Taoki, float-sand and float-stone must have relations via crafting with rest of the world
19:43 celeron55 VanessaE: and?
19:43 Taoki Heh. Cuz Minetest needs hardware lighting like christians need Jesus :)
19:43 VanessaE PilzAdam: how much of a region did you test?
19:43 PilzAdam the problem is the environment step after 0.1 seconds
19:43 Taoki Updating the lightmap like that is hacky IMO
19:44 VanessaE PilzAdam: exactly
19:44 Taoki Just one thing that's hacky about the current lighting
19:44 VanessaE the actual command executes just fine, but the server hangs like hell when it's done.
19:44 Taoki proller: Well normal dirt or stone can't be crafted, they're only natural
19:44 PilzAdam VanessaE, easy to fix: set minetest.timer_to_add = {} after looping through the dignodes in worldedit
19:44 celeron55 Taoki: i think all hardware lighting except raytracing is just full of hacks, this voxel lighting is good
19:44 VanessaE PilzAdam: no, WE DON'T BREAK USERSPACE.
19:44 proller Taoki, player must be need in float-* materials in some purposes
19:45 PilzAdam this doesnt break anything
19:45 VanessaE fix the problem in the engine code or in falling.lua
19:45 celeron55 Taoki: your intention seems mostly to be removing voxel lighting from minetest and replacing it with something that can only run on modern hardware, which i frown upon
19:45 VanessaE I ripped your change out and it fixed the problem.  Which means you need to only be adding those timers to FALLING nodes.
19:45 VanessaE not to ALL nodes that are dug
19:45 Taoki proller: Well we have sand and desert-sand currently, which are the exact same thing only different color
19:45 celeron55 or alternatively making it look like shit on older hardware
19:45 PilzAdam it needs to be done if a mod calls dig_node() excessively
19:45 Taoki celeron55: Not removing entirely as long as older hardware -needs- it
19:46 VanessaE PilzAdam: and what about all the mods out there that have to call nodeupdate()?
19:46 celeron55 Taoki: if it's not removed, then issues with it aren't removed either; thus, it is not a solution to anything
19:46 PilzAdam how much do they call it?
19:46 proller Taoki, better for some thing player must get some rare material from sky and from hell, it must encourage trade and world explore
19:46 Taoki celeron55: If everything we wanted was possible, I do wish it could be removed entirely. But that depends if HW lights would work with older hardware too. If not then it should stay optional
19:46 celeron55 Taoki: so don't propose it as any kind of solution
19:46 PilzAdam I dont think they call it for 10000+ nodes
19:46 VanessaE PilzAdam: 83 separate calls in my entire mods tree
19:46 Taoki Yeah I guess
19:47 VanessaE (and none of them are my mods)
19:47 PilzAdam VanessaE, for how many nodes per environment step=
19:47 PilzAdam *?
19:47 Taoki But I'll correct; Only removed as long as older hardware can't handle normal lighting. As in light entities from Irrlicht. Normally, those can be handled even without shaders... but like we discussed shades are needed to mask it in caves
19:47 Taoki So if the problem is still there then yeah
19:47 VanessaE PilzAdam: beats me, I have no way to count that up
19:47 RealBadAngel PilzAdam, introducing delay to stuff not connected in anyway to falling nodes is a bad idea
19:48 VanessaE I know technic makes significant use of it in constructors, deployers, node breakers
19:48 VanessaE I see lots of calls in mesecons
19:49 celeron55 Taoki: once the hardware we consider "old" runs such shaders reasonably, we'll probably throw the voxel lighting completely away; until then, it will stay
19:49 VanessaE hm, homedecor does make one call to it in the 3dforniture import (blame tonyka)
19:49 Taoki celeron55: Makes sense, that's correct. But if better lighting can be added, it should then stay optionally
19:49 PilzAdam RealBadAngel, why?
19:49 celeron55 Taoki: also, for that to happen at that time, there needs to be someone around who has or will implement the shadered shadows and stuff
19:50 celeron55 Taoki: currently, we have nobody who can do that
19:50 Taoki I know. I really wish I could, but I have virtually no diea about those things :(
19:50 Taoki Yeah, really sucks
19:50 PilzAdam VanessaE, every mod only calls it for single nodes
19:50 PilzAdam except worldedit
19:50 celeron55 also, those who can do that don't want to touch minetest; minetest is like total crap from their viewpoint
19:50 celeron55 they want stuff that is designed from the ground up for modern graphcis
19:50 Taoki They're fools :P
19:50 Taoki Oh, MineTest can have very modern graphics once this is done
19:51 VanessaE PilzAdam: still, you should put the delay in the FALLING code, not in nodeupdate.
19:51 PilzAdam nodeupated is the falling code
19:51 VanessaE I'm not saying to get rid of the idea, I'm saying to localize and limit its effects to that which you originally intended to fix anyway
19:51 VanessaE yeah and DIG NODE uses it!
19:51 celeron55 i could consider throwing some coffee money out of donation budget to someone who'd do that, if nobody does it for free
19:51 VanessaE you literally affect EVERY mod that uses dig_node() for any reason at all
19:51 Taoki celeron55: If shaders are used for light shadows, are the lights done there too? I keep hoping we can use the light entities in Irrlicht and speficy them in C++. ButI assume it would be a combination between that and shaders
19:51 RealBadAngel mesecons are falling stuff?
19:51 PilzAdam there is no way to have the delay when a player digs the node, and not have it when a mod calls it
19:52 VanessaE I'm just saying to move it up one level, out of nodeupdate() and into the code that calls nodeupdate.
19:52 VanessaE RealBadAngel: probably for a piston pushing objects around, or the movestone
19:52 Taoki celeron55: I'd donate to someone to do that too. Though we're sort of in a money crisis so even 10$ would have me killed :P But I'd certainly try it
19:52 VanessaE PilzAdam: then.  put. it. in. the. falling. code. itself.
19:52 PilzAdam VanessaE, not possible, since the player call dig_node() to dig nodes
19:52 PilzAdam what do you mean by falling code?
19:52 VanessaE where you convert each node into an entity
19:52 celeron55 Taoki: if you find some people with some odd fetishes to retro opengl wrapped in a clumsy cross-platform library (irrlicht), tell them minetest could use some work 8D
19:52 VanessaE that's the proper place to put it
19:53 Taoki celeron55: But 10$ for something like this could upset you if it's too little ;)
19:53 Taoki (cuz I don't see who else can at all)
19:53 VanessaE gotta run, back later.
19:53 Taoki later
19:53 PilzAdam <VanessaE> where you convert each node into an entity <- why havent you said this before?
19:54 VanessaE because I thought you would understand what I meant?
19:54 VanessaE I didn't think it needed to be stated that explicitly.
19:54 VanessaE anyway, out ->
19:55 PilzAdam this would cause a delay for the node at the postion of nodeupdate too
19:55 celeron55 PilzAdam: that is the original problem here?
19:55 celeron55 what*
19:56 PilzAdam a large table being generate/copied in the env step after calling nodeupate() for thousands of nodes at the same time
19:56 celeron55 that is the resulting problem of the current fix to the problem that i am asking about
19:57 PilzAdam the table has #(nodes)*8 elements
19:57 PilzAdam the original intention of me for adding the dealy was 1) entities arent spawned all at the same time and 2) it looks good
19:58 PilzAdam *delay
19:58 Taoki Anyway, something just as important: http://forum.minetest.net/viewtopic.php?pid=81733 I'm curious what people think about this especially the main devs
19:58 celeron55 why don't you just limit the size of the table
19:58 celeron55 if it grows higher than like 10, start just updating immediately
19:59 PilzAdam it could be fixed by moving the delay to the place were the entity is actually genereated
19:59 PilzAdam because the delay causes the table to grow
20:01 celeron55 i don't know... but if the performance is decreased 100x like VanessaE says, it isn't good
20:02 celeron55 there needs to be a way to make it efficient, even while cases that actually make it take seconds are silly
20:02 PilzAdam yep, fixed it
20:05 Taoki celeron55: BTW, regarding the voxel lighting. My own idea would actually still require it. Which is to use the current "lightmaps" as a way to limit hardware lights outdoor, at least for sun
20:05 Taoki So basically: When HW lighting is off, voxel light areas are used like they are currently; bright spots
20:05 Taoki When HW lighting is on however, they're used to indicate where the infinite light can shine
20:06 Taoki It would also remain needed so the light value at each node can be detected, which is needed by mods. For example to know when it's dark so you can spawn a mob
20:07 Taoki The only real issue is how to use this as a mask for lights. That I can't tell
20:08 celeron55 i can imagine tricks that don't even need shaders, but they're hard and might need bare opengl (but i've said this before)
20:08 Taoki We have the current voxel light system which knows what light value each node should have from the sun / moon, and we can add an Irrlicht light with just one line of code. But how to combine them is the thing :P
20:08 Taoki yeah, those tricks would be awesome
20:10 Taoki There actually is a trick that can be used currently as well; Add a black texture over nodes to cancel lighting or dim it. Using the existing light calculations of the voxel system
20:11 Taoki Or fill color that covers the texture. I assume multi-texturing likely works under Irrlicht
20:11 Taoki Pretty sure you can set surfaces to not be affected by lights, or limit some surfaces to some lights only.
20:12 Taoki Problem is how to do it smoothly AND not break lights coming from torches placed indoor
20:14 celeron55 well, you can always do multiple render passes without clearing the frame or z buffers (and you should be able to swap lights between them)
20:14 celeron55 that'll more or less eat framerate though
20:14 Taoki Yeah. That might be doable without shaders too
20:14 Taoki hmm
20:15 celeron55 and isn't trivial anyway
20:18 celeron55 one could render all kinds of special masks based on what receives voxel sunlight or light at all or something, and then AND or OR that to something else and whatever
20:18 celeron55 but it's again very nontrivial
20:18 celeron55 especially getting something useful out of it
20:18 Taoki Having real light is an important thing I believe. I think it's the most major feature that would be needed
20:19 Taoki There are others too but especially this
20:19 Taoki I think special masks are sort of the best or only way to go though
20:19 hmmmm [01:38 PM] <proller> ^^^ <proller> and springs make not very good lakes in air above unloaded blocks
20:20 celeron55 i think it would be reasonable to only implement it with shaders
20:20 hmmmm yeah, that's something i couldn't really figure out when i first saw the problem
20:20 celeron55 or maybe not
20:20 celeron55 it's hard to decide where to draw the line
20:20 hmmmm i don't know, i need to look at it more, but either way, the insane SEnv lag when water flows into an unloaded chunk == unacceptable
20:20 Taoki Shaders are usually for improved effects anbd the like. Lighting would be... sorta too important to be shader-only
20:21 hmmmm need to add some sort of check in updateLighting() so it doesn't bother to keep trying to update that
20:21 celeron55 in modern 3d graphics, you don't have fixed pipeline at all and use shaders for everything, including lights
20:21 hmmmm [02:25 PM] <celeron55> 0.3 is like the easiest possible, 0.4 is like the hardest possible
20:22 hmmmm all 0.3 versions have problems with missing libintl for me
20:22 hmmmm i need to remove -lintl from a certain cmake file to get anything older than 0.4.1 to build
20:22 Taoki heh. Way too modern :P
20:22 hmmmm 0.4 is hard because of the JThread mess and the json crap
20:22 proller hmmmm, very simple to fix water in transform_liquids - drop if bottom unloaded
20:22 celeron55 Taoki: that's how you'd do any new project if you started it today
20:22 hmmmm proller, need to make it more general
20:23 hmmmm i'll look at it more later on
20:23 hmmmm first i need to go pass out
20:23 Taoki celeron55: Thing is that Irrlicht light entities should also work for that old hardware which can't handle shaders. And barely reduce performance if at all
20:23 celeron55 Taoki: true
20:24 hmmmm i couldn't play eduke32 with dynamic lighting even
20:24 hmmmm and i don't think an 8400gs is too old
20:24 Taoki And there are some mask hacks I can think of to hide sunlight in caves. At worst we use a black color on the texture, opposite of the current voxel lights. Where the suinlight doesn't shine at all, the surface would be fully black. Black can't be lit
20:25 celeron55 but really, this problem needs to be thought from the standpoint of using hardware lighting, and then asking the question: how can voxel lighting be used to make caves dark, *without* making torches in them dark
20:25 Taoki And to make sure that unlit caves aren't 100% darkl unless a torch is placed, bumped up manually by some ambient lighting
20:25 proller hmmmm, other variants will be much more difficult, or impossible  if you make infinity spring at 30000
20:25 Taoki celeron55: Yes. Thgat's how I'm thinking of it
20:26 Taoki The closest hack in mind though is using black color to kill sunlight. But that would also kill torch lights
20:26 proller hmmmm, best way is load block under, but you cant load all blocks to -31000
20:27 Taoki I feel we're pretty close with a solution though... somehow. Still far but close :P
20:27 celeron55 it's just that you haven't yet thought about the thing that will make the solution not work at all
20:28 Taoki yeah. Torches would now be the problem
20:28 celeron55 but really, one would first need to know all the things that can be done with irrlicht with old hardware
20:28 celeron55 it's probably more than one would think, and probably different things than one would think
20:31 celeron55 hmm, assuming it is possible to render a "mask" that can be used to dim everything not touched by sunlight, it would be possible to just let it dim torches when you're in sunlight, and when you go to a cave and don't look outside, the dimming would be turned off and ambient light would be turned off
20:31 hmmmm proller, can the springs spawn above water_level?
20:31 celeron55 -> torches are bright
20:31 Taoki Ahhh, I think I remembered something actually. If I'm correct there is a diffuse parameter for faces in Irrlicht. It allows changing how much light a surface takes, without changing its color. I hope I'm right about that
20:32 Taoki You can adjust diffuse and specular separately yeah... I think
20:32 celeron55 that would be a reasonable solution and would need only single extra pass to make the mask
20:32 Taoki Yes. Can someone else check just to be 100% sure?
20:32 celeron55 (what i said, i mean)
20:32 Taoki Ah, ok
20:32 Taoki I think I saw the diffuse option when I implemented my models
20:33 Taoki Oh, I remember better. It's called "diffuse color". I already added it to colorize entities. But if all 3 values are adjusted together than it just decreases how much lighting the surface takes
20:33 Taoki diffuse_color, specular_color... I think there was a third one but mattered less
20:34 celeron55 minetest basically currently uses diffuse color to set the lighting of *everything*
20:34 celeron55 afaik
20:34 celeron55 you can't really use diffuse color to set lighting if hardware lighting is enabled
20:34 celeron55 it's very nonlinear and unpredictable then
20:34 Taoki Found my lines: m_animated_meshnode->getMaterial(i).AmbientColor = m_prop.colors[i]; m_animated_meshnode->getMaterial(i).DiffuseColor = m_prop.colors[i]; m_animated_meshnode->getMaterial(i).SpecularColor = m_prop.colors[i];
20:35 Taoki So there's AmbientColor, DiffuseColor, and SpecularColor
20:35 Taoki I wonder what the difference is between AmbientColor and DiffuseColor
20:35 celeron55 http://blog.lexique-du-net.com/index.php?post/2009/07/24/AmbientDiffuseEmissive-and-specular-colorSome-examples
20:36 PilzAdam celeron55, VanessaE, here is a working fix: https://github.com/PilzAdam/minetest/commit/97f0bb03423b6d2e22058166b677e568c53d7567
20:36 Taoki Oh... I see
20:36 celeron55 that is, diffuse color will show up when you point it with non-ambient light and ambient color when it isn't pointed with such.... ehm... well, i can't say i would understand how that would actually work
20:37 proller hmmmm, if you write it to register_ore with height >0 and param128, i tried it on my server, its ok.  and its only way to run down witout death
20:37 Taoki Me neither. Sounds more like bounce lighting
20:38 Taoki Still, this parameter does bring us closer to a solution hopefully. Only issue would be how to separate sunlight from torches and torches from each other as well
20:39 Taoki Hopefully all with Irrlicht functions
20:40 celeron55 i don't think that is close to a solution
20:49 Taoki celeron55: Someone in #icclicht mentioned a light manager, which could solve our problem. Supposedly it can separate nodes from certain lights
20:50 Taoki Found a video which uses it, the description is interesting: http://www.youtube.com/watch?v=RO3pRcymG-w
20:50 Taoki He also linked this: http://irrlicht.sourceforge.net/docu/example020.html
20:51 Taoki "Normally, you are limited to 8 dynamic lights per scene: this is a hardware limit. If you want to use more dynamic lights in your scene, then you can register an optional light manager that allows you to to turn lights on and off at specific point during rendering. You are still limited to 8 lights, but the limit is per scene node."
20:52 celeron55 that doesn't help at all in our problem
20:53 Taoki Ahh. Got the impression it's exactly it. Since it allows separating lights per nodes
20:53 celeron55 the only thing it does is multiplexes light sources to allow having more than 8
20:54 Taoki ah
20:54 Taoki He still said the light manager could be used to achieve this. Maybe not in that example
20:54 celeron55 i am wondering that you might have even asked
20:55 Taoki [23:42:31] <Taoki> Separate question, not sure if I asked some months ago already: Is there really no way to make some surfaces be affected by some lights only? Some function, or hack...
20:55 Taoki [23:46:01] <Taoki> Can be multiple nodes too at worst. But for example if I place two surfaces one next to the other and two lights one next to the other. I mean some way to make one surface affected by one light and the other by the other light
20:55 Taoki Hold on actually,
20:56 Taoki http://pastebin.com/cmVBWtX3
20:56 celeron55 that is simply achieved by swapping lights on and off between rendering different stuff; there's no magic there
20:56 Taoki No magic, but would it work?
20:57 celeron55 are you aware that node means something completely different in irrlicht than in minetest?
20:57 Taoki I think. A node in MT is usually a block. Nodes in Irrlicht are like holders for meshes IIRC
20:58 celeron55 something like that
20:58 Taoki Still, I think each surface of a block in MT is an Irrlicht node. Or at least each whole block
20:58 Taoki From what I quickly seen in the code once
20:59 celeron55 actually the whole minetest voxel world is a single irrlicht node
20:59 celeron55 but it's irrelevant here (except that it makes the direct use of the light manager not possible)
20:59 Taoki Right. In that case there is a problem
20:59 Taoki ok then. Was going to ask if separating caves into other nodes could allow them to be lit separately
21:00 celeron55 we can switch lights on and off in the map rendering code anyway
21:00 celeron55 or, well, world
21:00 celeron55 Taoki: i believe so
21:00 Taoki True. But once could be seeing multiple lights at once
21:00 Taoki Both sunlight and a torch in a cave
21:00 Taoki **one
21:01 celeron55 basically, stuff designated as being caves or insides of things would be rendered without the sunlight light
21:01 Taoki Yeah. Caves could be separated into a second Irrlicht node with its own lighting effects
21:01 celeron55 no
21:02 celeron55 don't mix up any irrlicht nodes in here
21:02 Taoki ok
21:02 celeron55 it's irrelevant here and only makes things hard
21:02 Taoki Hmm... and there's another problem in the mix; Torches themselves would need their lighting cut. Since you wouldn't want to place a torch on one side of the wall and have its light on the other... even in a deep cave with no sun light
21:02 RealBadAngel http://irrlicht.sourceforge.net/docu/example003.html
21:03 celeron55 Taoki: for that and for any shadows outside you need shadows; there are someting like two ways to make them traditionally using irrlicht
21:04 celeron55 they require two render passes and whatever
21:04 Taoki Yeah, I think the same
21:04 RealBadAngel the sample code of scene node looks suspiciously similar to mt node ;)
21:04 Taoki Well, we can go with that if it's really really needed. How much can two passes decrease performance
21:05 celeron55 RealBadAngel: IIRC minetest started as that example
21:05 celeron55 altough all custom scene nodes always are implemented like how minetest implements the world
21:06 Taoki There is another way too. But eeven more FPS costly
21:06 RealBadAngel so, we *can* treat mt node as irrlicht scene node object
21:06 Taoki That would be to use dynamic shadows / light cutting. So every surface casts a direct sharp shadow. But that's usually meant to be an option itself cuz it's pretty consuming to do for each block
21:06 celeron55 Taoki: well, 2x
21:07 Taoki Only issue with light cutting would be if the ceiling of a cave hasn't loaded (eg: is beyond your draw distance). No blocks are detected and the system could think you're outside. But I guess that's less important for starters
21:07 Taoki Do Irrlicht lights support cutting natively?
21:08 celeron55 it would be quite deceptive to try to think that two passes would take less than twice the time 8)
21:08 Taoki heh
21:08 celeron55 but if you take into account more things, then it will become (much) less
21:08 sapier1 joined #minetest-dev
21:08 celeron55 but that depends on what you do with two passes
21:09 celeron55 if you're drawing shadows, then you could probably render a much smaller distance for the shadow pass
21:09 Taoki yeah. And exclude other defails perhaps
21:10 celeron55 there's not really any such details in minetest
21:11 celeron55 anyway... somebody'd just need to start experimenting
21:12 Taoki celeron55: How much worse would dynamic shadows be?
21:12 celeron55 dunno
21:13 celeron55 3d shadows are standard stuff, just google
21:13 Taoki That sounds like it could be quite the solution
21:13 Taoki FPS costly, but better than nothing. Especially if HW lighting can be optional
21:14 Taoki Haha. The person who I spoke to about this is actually on this channel too. Hi serengeor :)
21:15 Taoki celeron55: But yeah. From what I can think, this would be the best if not only solution. And really worth trying out :) Dynamic shadows would instantly do the trick. But I wonder how reliable they are for that...
21:16 Taoki Irrlicht should support it natively and without shaders
21:16 celeron55 ehm, how do you think "dynamic shadows" differ from "shadows"?
21:16 Taoki I mean the same thing sorry
21:17 celeron55 irrlicht can do it, but it's not automatic by any means; it's basically the same as doing it with bare opengl/d3d
21:17 Taoki Hopefully that's ok then
21:19 Jordach can i have a main page suggestion?
21:19 Taoki As in, still doable with Irrlicht
21:19 Jordach as in: http://minetest.net
21:21 Taoki sure
21:21 Jordach celeron55, instead of linking directly to the engine github repo, instead, is it possible to link it to https://github.com/minetest ?
21:21 Taoki Sounds better indeed
21:36 serengeor Taoki, what kind of lighting exists currently in minetest?
21:36 Taoki serengeor: A fake voxel lighting system. Basically, light value is calculated at various points, and the surface's color is changed to reflect it
21:37 Taoki Long story short, light is done by fakely coloring each surface or vertice. But there's many reasons to switch to hardware lights
21:38 Taoki However, if that's done we need a way to keep sunlight from shining in caves. The only info the current light system offers IRC is "the light value at each block". Probably a number between 1 and 10 or something like that.
21:38 Taoki So that info needs to be worked with to mask different lights for different areas
21:41 serengeor I see
21:42 Taoki serengeor: But the situation is complicated. Since for example you can place a lot of torches at once in a cave, and their lights need to show correctly. You might also be seeing the sunlight outside at the same time
21:42 Taoki What would make everything right was if Irrlicht had a way to influnce the effect a specific light has on a specific surface
21:42 Taoki And tell each face or vertice how much light to take from any individual light entity
21:43 serengeor you can use vertex colors to do that
21:44 serengeor and use shader to mix light with vertex color
21:45 serengeor but that just sounds the same as it is now :?
21:45 Taoki serengeor: Problem is that MineTest is meant to run on machines that don't support shaders too. So it would have been awesome to have a shader-less way to do this as well
21:45 Taoki Kind of. It's better to mask a real hardware light than manually change the color of surfaces
21:45 dexter0 3d textures, oh wait.
21:46 Taoki serengeor: Can you think of a shader-less way to do this in Irrlicht though?
21:47 serengeor raytrace the fuck out of it
21:47 Taoki Even if it requires a hack to hide lights in some places... like for instance colorizing that face in black somehow
21:47 serengeor :D
21:47 Taoki heh, yeah that would be one idea :P
21:48 Taoki I suggested dynamic shadows, I think they might be the best option
21:49 dexter0 Unless you use stencil shadows, you'd need to build a shadow map for each light.
21:49 dexter0 err shadow volumes
21:51 Taoki Yep. Stencil shadows would be the way
21:52 Taoki But how do you place a light with stencil shadowing?
22:17 proller joined #minetest-dev
22:19 iqualfragile joined #minetest-dev
22:20 Taoki Ah-hah! This shows how to add shadows to lights... hopefully this is what we need: http://irrlicht.sourceforge.net/docu/example008.html
22:33 VanessaE PilzAdam: that seems to work exactly as desired.
22:33 VanessaE worldedit is fast again, and a large region of e.g. falling sand falls in a nice, even wave (until the client starts to lose fps ;) )
22:46 Deivan joined #minetest-dev
23:05 hmmmm i don't mean to be insulting, but gosh, the forum is filled with what seems like 12 year olds
23:05 hmmmm just read the Moontest thread, pretty lol
23:06 PilzAdam maybe because only 12 year old people play Minetest?
23:06 hmmmm :(
23:06 Taoki Far from it
23:07 * Taoki is a 12 * 2 year old :P
23:17 Taoki BTW. I found out a new method of trolling in MineTest which just happened to me on a server. Trolls can dig holes in stone. If a new player falls in them without either a pickaxe or nodes to put under themselves, and there's no command to teleport to spawn or otherwise, you're stuck there
23:19 Taoki The server's fault for having neither /spawn, /die, or default privileges for either free_move or /teleport
23:31 ShadowNinja I think lua should be able to get a spawn point for servers that don't have static_spawnpoint set

| Channels | #minetest-dev index | Today | | Google Search | Plaintext