Minetest logo

IRC log for #minetest, 2017-11-03

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

All times shown according to UTC.

Time Nick Message
00:00 sovetskiy thanks
00:02 shymega joined #minetest
00:04 o11c sovetskiy: if you really want to use all your own mods, at least borrow bits of their code - open source is a thing for a reason
00:08 sovetskiy I don't want to share my code. Relax. My mods for me only
00:09 EvergreenTree joined #minetest
00:17 nowhere_man joined #minetest
00:32 lumberJ joined #minetest
00:38 Sokomine joined #minetest
00:42 EvergreenTree joined #minetest
00:55 wilkgr joined #minetest
00:57 orbea concerning that meshnode crash, we have a reliable test world to reproduce the crash now. https://github.com/stujones11/meshn​ode/issues/6#issuecomment-341600300
00:57 orbea crashes every time maybe 30 seconds
00:57 orbea maybe amount of distance traveled?
01:18 EvergreenTree joined #minetest
01:22 Tmanyo joined #minetest
01:22 ssieb joined #minetest
02:00 orbea backtrace of meshnode crash with git version https://pastebin.com/j1HUq5aT
02:00 orbea https://github.com/stujones11/meshn​ode/issues/6#issuecomment-341606980
02:17 Tux[Qyou] joined #minetest
02:27 bigfoot547 Guys, I did a sudo make install with minetest-0.5.0-dev
02:27 bigfoot547 And I compiled minetest-0.4.16
02:28 bigfoot547 I did sudo make install and it told me "Up-to-date" on all the files
02:29 Calinou bigfoot547: "make clean" to clean all build files
02:29 Calinou do this before switching versions, or to ensure a clean build
02:29 bigfoot547 Ok
02:29 bigfoot547 Do I need to do sudo make clean?
02:29 Calinou to force recreation of CMake files: rm CMakeCache.txt
02:29 Calinou no
02:29 bigfoot547 Cuz only root has access to /usr/local/share
02:29 bigfoot547 maybe
02:32 bigfoot547 Well, I did make clean
02:33 bigfoot547 then sudo make install in the stable build
02:33 bigfoot547 But it just said "Up-to-date" on all the files
02:34 bigfoot547 And minetest --version says 0.5.0-dev
02:39 bigfoot547 Calinou?
02:52 XeonSquared joined #minetest
03:10 Hawk777 joined #minetest
03:10 wilkgr joined #minetest
03:11 Billre left #minetest
04:15 AlexYst joined #minetest
04:24 AlexYst joined #minetest
05:08 soloojos joined #minetest
06:40 xerox123 joined #minetest
06:55 silent_elk joined #minetest
07:02 frostsnow joined #minetest
07:07 wilkgr joined #minetest
08:25 Telesight joined #minetest
08:27 nowhere_man joined #minetest
08:32 Darcidride joined #minetest
08:39 troller joined #minetest
08:50 dabbill joined #minetest
08:52 YuGiOhJCJ joined #minetest
08:53 ensonic joined #minetest
08:55 Fritigern joined #minetest
09:27 ensonic joined #minetest
09:35 troller joined #minetest
09:41 asl97 joined #minetest
09:43 Jousway joined #minetest
10:18 nowhere_man joined #minetest
10:23 troller joined #minetest
10:25 YuGiOhJCJ joined #minetest
10:35 Fixer joined #minetest
10:41 troller joined #minetest
11:07 troller joined #minetest
11:14 nac joined #minetest
11:20 NathanS21 joined #minetest
11:28 Krock joined #minetest
11:33 nowhere_man joined #minetest
11:37 CWz joined #minetest
11:39 deltasquared joined #minetest
11:48 Fritigern joined #minetest
12:18 Calinou bigfoot547: you need to recompile Minetest using "make" after doing "make clean"
12:18 Calinou although "sudo make install" compiles if needed, IIRC
12:18 Calinou (but it compiles as root, which isn't great for security and permission handling)
12:19 deltasquared pretty sure the standard pattern is "make && sudo make install"
12:21 EvergreenTree joined #minetest
12:28 Wuzzy joined #minetest
12:30 Calinou yeah
12:30 Calinou or "make -j$(nproc) && sudo make install" :P
12:33 deltasquared Calinou: hmm, I'm used to build systems that can sanely handle multi-threading safely by default
12:34 deltasquared make can be easy to get dependencies wrong with
12:45 Jordach joined #minetest
12:53 ensonic joined #minetest
13:04 RobbieF joined #minetest
13:04 RobbieF joined #minetest
13:05 RobbieF left #minetest
13:24 Billre joined #minetest
13:35 antims joined #minetest
14:20 deltasquared joined #minetest
14:25 Enricoo joined #minetest
14:26 Telesight joined #minetest
14:29 Enricoo joined #minetest
14:35 CalebDavis joined #minetest
14:37 Tux[Qyou] joined #minetest
15:28 twoelk joined #minetest
15:36 nac joined #minetest
15:36 SanskritFritz joined #minetest
15:45 proller joined #minetest
16:06 paramat joined #minetest
16:17 proller joined #minetest
16:28 MinetestBot [git] paramat -> minetest/minetest_game: Binoculars: Only set property if necessary to reduce mesh reloading 08c60cf https://git.io/vFcet (2017-11-03T16:26:00Z)
16:28 MinetestBot [git] paramat -> minetest/minetest_game: Player model: Remove vertical stretch of body and limbs, remove cape 9596d00 https://git.io/vFceq (2017-11-03T16:25:54Z)
16:39 jas_ i'd love to be able to crouch in minetest.  that'd be sweet.
16:42 nowhere_man joined #minetest
16:48 coffee joined #minetest
16:48 coffee left #minetest
16:50 H4mlet joined #minetest
16:59 nowhereman joined #minetest
17:19 QwertyDragon joined #minetest
17:22 troller joined #minetest
17:48 GreenDimond joined #minetest
17:49 GreenDimond xpanes.register_pane seems to ignore `paramtype = "light",` and `light_source = bla`
17:55 GreenDimond is there any way to get it to work?
18:00 Krock modify xpanes
18:01 Krock or overwrite the node definition after calling register_pane
18:01 GreenDimond how do I overwrite the def?
18:03 Krock in your mod
18:03 GreenDimond I know, but what function do I use?
18:03 Krock minetest.override_item("node name", { new def})
18:04 Krock https://github.com/minetest/minetes​t/blob/master/doc/lua_api.txt#L2429
18:04 Krock !next
18:04 MinetestBot Another satisfied customer. Next!
18:04 GreenDimond Ah, thankyou.
18:04 GreenDimond I was searching the api for `overwrite` instead of `override` xD
18:14 ensonic joined #minetest
18:30 paramat joined #minetest
18:42 troller joined #minetest
18:57 ssieb joined #minetest
19:13 sonicpp joined #minetest
19:14 GreenDimond joined #minetest
19:15 GreenDimond So, I am trying to use `^[transformFX` on a pane texture, but it doesnt seem to work. Using it in an override doesnt work either.
19:17 deltasquared joined #minetest
19:20 Krock is there any code that could be pasted?
19:21 fwhcat joined #minetest
19:22 johnnyjoy joined #minetest
19:22 GreenDimond not that would help..
19:23 Krock sometimes it can. Are you sure .override_item is called during the mod loading
19:23 Krock *?
19:23 GreenDimond Yes, because the other things I override work. The reason the override doesnt work is it doesnt like def.texture
19:24 GreenDimond I figured out I had to define def first
19:24 GreenDimond but then it still wouldnt take the texture
19:28 Krock minetest.override_item("mod:node", { tiles = "default:dirt^[transformFX" })  should work just fine o.o
19:28 GreenDimond Krock: Here is the code https://gist.github.com/GreenXenit​h/d7ad9ccc1c82a4bdd51e01a18140bd64
19:28 GreenDimond well theres a problem with that
19:28 GreenDimond the pane function changes tiles to def.textures
19:28 GreenDimond so `tiles = {def.textures[3], def.textures[3], def.textures[1]},`
19:30 GreenDimond overriding that in the def doesnt work if I use tiles for some reason
19:33 Krock why don't you override "xpanes:lights_"..nodename.."_pane_flat" ?
19:34 Krock is that not an issue?
19:34 RedSoxFan07 joined #minetest
19:39 LazyJ joined #minetest
19:44 GreenDimond Krock: Sorry, was talking to someone. I do override the flat one. Or... wait. No I dont :o That might be helpful..
19:45 Krock Don't worry. IRC is a place where answers can take hours
19:46 Fixer days
19:46 Fixer or even eternity
19:48 deltasquared is there a way I can get MT to more strictly control the memory it uses? it's getting killed from OOM related errors currently just by wandering around on a server.
19:48 deltasquared it appears to be keeping the world in RAM and not evicting old chunks.
19:48 deltasquared (dunno if local world saving would alleviate this)
19:49 Krock remove luajit or ensure it's a x64 build
19:49 deltasquared it's a 64-bit build, removing luajit would not make a difference when it's client only methinks
19:49 deltasquared this is like 3.5GB we're talking here when it OOMs, nearly all of my system RAM
19:49 Krock ah I see. Client non-luaJIT OOM - one of the less popular cases
19:50 deltasquared so, it's known about?
19:50 Krock also happens to me when using a 2048px texture pack
19:50 Krock you can reduce the amount of cached blocks client-side
19:50 deltasquared Krock: oh yeah?
19:51 deltasquared *looks at advanced options*
19:51 Krock and also a lower texture pack resolution, which helps quite much
19:51 GreenDimond Krock: Overriding the _flat pane to set the tiles makes the front faces blank and gives the top a texture (both of which I dont want)
19:51 deltasquared the fact that I have to set that manually kinda sucks though.
19:51 deltasquared Krock: I'm only on 16x textures here
19:51 Krock GreenDimond, at least something happens
19:51 GreenDimond Yeah but it makes no sense
19:51 KrimZon_2 joined #minetest
19:52 Krock deltasquared, okay, that's strange. Are there any heavy mods like mobs or music-playing ones?
19:52 deltasquared Krock: no, this is a lightweight sever
19:52 deltasquared *server.
19:52 deltasquared VE-building, to be precise
19:52 deltasquared it definitely happens as I  explore the world more
19:52 Krock *checks whether there are VE-lightweight servers*
19:53 Wuzzy "Minetest: Because blocks want to be free."
19:53 deltasquared RAM is reasonable when I initially join and the world loads, but it only increases from there as I wander
19:53 deltasquared Krock: point is, what's that option called, I just want MT to kick out old chunks when they're not visible
19:53 deltasquared I don't think it's a memleak somewhere else as it doesn't continue to increase when I stand still.
19:54 deltasquared ... aha, client_mapblock_limit?
19:54 Krock https://github.com/minetest/minetest/b​lob/master/minetest.conf.example#L1011
19:54 deltasquared it's set to 5000, ouch
19:55 Krock see also: client_unload_unused_data_timeout = 600
19:55 deltasquared I'm suprised these aren't set to something more conservative by default
19:55 GreenDimond Because by default most devices can handle it
19:55 Krock 10 minutes and 5000 blocks are hard to reach
19:55 paramat 5000 is not a big volume
19:56 paramat it's only 266 nodes ^ 3
19:56 paramat so what you see with view range 133
19:56 paramat roughly
19:57 paramat the amount of RAM 500 blocks uses depends on mesh complexity
19:57 deltasquared paramat: mapblock suggests chunks though
19:57 paramat *5000
19:58 paramat no
19:58 paramat mapchunk is not a mapblock
19:58 deltasquared huh, I was under the impression that mapblock meant chunks in the engine
19:58 paramat block = 16 ^ 3 nodes
19:58 paramat chunk = 5 ^ 3 blocks by default
19:59 deltasquared paramat: ah, I had always used chunks in the MC sense
19:59 paramat argh
19:59 deltasquared so I guess I *should* call them mapblock just to to be sure
20:00 Krock <paramat> it's only 266 nodes ^ 3
20:00 Krock moment.. I have to do the math
20:00 deltasquared interesting, I'm setting it to 1024 and the timeout to 30 and now I'm getting segfaults not long after joining a server
20:01 Krock yes, matches quite well. It's just that servers only rarely send blocks that far away
20:04 paramat timeout too low is bad, unnecessary resends of blocks
20:04 paramat maybe 300 5mins
20:05 deltasquared paramat, right fine. still getting interesting crashes now though
20:05 GreenDimond Krock: So, what do you figure I should do?
20:07 Krock GreenDimond, uh.. let me check it myself
20:07 deltasquared ok wat, now I'm getting zlib deflate errors?
20:09 deltasquared welp now I'm getting segfaults all over the place
20:11 deltasquared “compressZlib: deflateInit failed”
20:11 deltasquared ...
20:14 sy sounds fun!
20:15 deltasquared I find it hard to believe all this weirdness started just as a result of disabling overcommit on my system
20:15 deltasquared there's still memory *free* when this stuff happens
20:16 Krock GreenDimond, https://pastebin.com/raw/WitZiLKS .. works for me. It's shiny and the torch is animated
20:17 deltasquared oh yeah, that's another thing, minetest doesn't let go of the memory it takes up while in-game - go on VE-building, goes up to around 1.5GB. quit and go to a singlenode test world, stays there - but loading that same map from a fresh start only takes up 0.8GB
20:19 deltasquared (includes other processes running but the relative gap should be apparent)
20:19 Krock roughly 300 MiB use in my singleplayer world
20:20 twoelk deltasquared: are you saving rto local at the same time?
20:20 twoelk -r
20:22 * twoelk decides to go do some mining
20:30 soloojos joined #minetest
20:30 deltasquared twoelk: no I am not currently
20:30 deltasquared ... or did I leave that on, hold on
20:31 deltasquared would that help or hinder in this sitation
20:32 deltasquared (it had been left on, just gone and turned off)
20:32 twoelk any difference in performance?
20:34 deltasquared twoelk: still testing, it's not really performance that's the issue
20:34 deltasquared the random crashes are :P
20:34 twoelk yeah sloopy wording on my side
20:36 deltasquared that does seem to have helpd a bit though
20:36 twoelk my client sometimes freezes when writting to localdatabase and sometimes crashes if low on memory - might be related or not
20:36 deltasquared OOM during DB flush... yeah I could see why that'd be a bit problematic
20:37 deltasquared .... aaaand again.
20:37 deltasquared it doesn't seem to like me moving around fast that's for sure
20:40 GreenDimond arg
20:40 GreenDimond why arent texture directions documented anywhere?
20:40 deltasquared good question, I was going to draw up like a texture with arrows on so I could figure it out
20:43 deltasquared hmm, does MT have any sort of minimum requirement specs list anywhere, I'm wondering if 4GB is considered pushing it
20:43 twoelk some like that have been created multiple times, maybe they should be part of minimal for use in testing
20:44 GreenDimond oh wait
20:44 GreenDimond it is docd
20:45 deltasquared GreenDimond: is it in lua_api.txt or elsewhere
20:45 GreenDimond yes, api
20:45 GreenDimond line 4499
20:45 deltasquared bladdy hell
20:46 deltasquared GreenDimond: erm, can you give me a phrase to search for, line numbers tend not to be reliable in my experience
20:47 paramat 4GB is fine, reducing client mapblock limit should limit RAM use
20:47 GreenDimond deltasquared: -Y
20:47 deltasquared GreenDimond: ... hrm, I was hoping for something that would indicate facedir directions for each one, but ok
20:48 deltasquared paramat: I did, I've set it all the way down at 256. still managing to crash in strange ways though
20:51 paramat afaik by far the largest RAM use for a client is the mapblock meshes
20:51 deltasquared that would figure
20:52 deltasquared I'm just puzzled by these segfaults that are occuring without so much as a peep on stderr
20:52 deltasquared I don't feel inclined to muck with a debug build of it at this time in the evening
20:52 o11c what's the backtrace at least?
20:53 paramat we don't have any client memory leak issues currently reported, afaik
20:53 deltasquared o11c: sometimes it doesn't have any whatsoever, like the stack was mangled
20:53 o11c run under valgrind? Might be too slow
20:54 o11c -fsanitize=address is much faster but requires recompiling
20:55 deltasquared o11c: like I said, I won't be trying anything like that this evening, I'm a bit tired now :P
20:56 deltasquared relevant part of the log is this - https://ptpb.pw/XU5T.txt and it really is just that one line in the stack trace
20:57 deltasquared ..oh wait hold on I missed a bit before it
20:58 deltasquared https://ptpb.pw/ouG6.txt there ya go
20:58 deltasquared crash is happening inside mesa? welp
21:00 deltasquared sometimes I wish I didn't have to recompile distro libs on arch just to get sane stack traces
21:01 GreenDimond Krock: So, I got the texture to flip, but if I place on pane next to another, the texture flips.
21:01 GreenDimond which makes it look weird
21:02 troller joined #minetest
21:04 Krock GreenDimond, then make it world-aligned
21:04 GreenDimond Isn't that a 0.5.0 thing?
21:04 Krock ^ requires a frsh 0.5.0 client
21:04 GreenDimond Yeah no can't do that
21:06 RedSoxFan08 joined #minetest
21:14 deltasquared joined #minetest
21:15 o11c deltasquared: distros should have -dbg or -dbgsym packages
21:17 RedSoxFan07 joined #minetest
21:18 deltasquared o11c: actually thinking about it, I'm thinking of the wrong thing as you'd still see symbols if the stack was valid, just wouldn't be able to see source code in the debugger
21:19 deltasquared ... then again, maybe not if not exported. argh.
21:19 deltasquared in any case, arch does *not* do sep. debug packages and it doesn't compile with -g or anything like that
21:20 deltasquared nor does it do -dev packages... all those little header files included by default hurts FS find performance that's for sure
21:20 sfan5 muh fs performance
21:20 sfan5 maybe don't install arch if you care about such things
21:20 deltasquared sfan5: it's never been an issue thus far...
21:21 sovetskiy joined #minetest
21:21 benrob0329 Void Linux FTW!
21:21 sovetskiy minetest uses Lua v5.1. Right?
21:21 GreenDimond Krock: Well it works :D so thanks :D
21:22 Krock sovetskiy, 5.1, LuaJIT or anything newer
21:22 Krock GreenDimond, ok, nice :)
21:23 sfan5 Krock: incorrect
21:23 sovetskiy so, there is no "continue" statement. right?
21:23 Krock why would one want to use a lower version if the mods might not work with it?
21:23 sfan5 highest supported is 5.1
21:24 Krock oh really? I didn't expect API changes on their side
21:24 deltasquared sovetskiy: nope, there's no "continue" keyword
21:24 deltasquared which annoys me quite a lot
21:24 sovetskiy damn!
21:24 sovetskiy continue statement is SO useful
21:24 ensonic joined #minetest
21:25 deltasquared I can't remember where it was I tried to use it, though I do recall having to be somewhat creative to work around it
21:25 Krock sovetskiy, there's the trick to use  do ... until false  (or was it do ... while false?) and break inside
21:25 sfan5 there's a few changes that would need backwards compat
21:25 sfan5 shadowninja has made prs for 5.2 and 5.3 compat but they were rejected
21:25 Krock oh great <.<
21:26 Krock well, at least this guarantees that mods will work with LuaJIT
21:27 deltasquared I for one would block allowing higher lua versions if it meant luaJIT wouldn't work, it's kinda useful
21:27 sovetskiy do...until -- loop inside loop only the sake of termination of current iteration? Idiotic.
21:29 Krock it's not about being helpful or not - it works, and actually saves some indents of if-checks
21:29 Krock s/helpful/idiotic/
21:30 sovetskiy what about simple "goto" then?
21:30 Krock that was added in a newer Lua version which is, what I just learnt, not supported yet
21:30 RedSoxFan07 joined #minetest
21:30 sovetskiy how long shoudld we wait?
21:31 Krock depends whether Mike Pall is going to change his opinion on adapting newer Lua features in his JIT version or not
21:36 sovetskiy because of lack of "continue" my code shifting right all the time
21:36 orbea hey Wuzzy I checked out the mana, teletool and returnmirror mods, pretty cool. Would you be willing to make the returnmirror and magic teletool craftable even if hidden behind a configuration setting?
21:37 Wuzzy no
21:37 orbea okay...
21:39 deltasquared hmm, /me suspects one misses the intent of a config variable if one asks for it to be bypassed
21:39 deltasquared IOW moan at your server operator about it :P
21:40 Krock sovetskiy, if there are more than two levels of indents, consider using the loop trick
21:40 orbea that is me, I just wanted to know if I could get this fixed in the mod before considering the implications of forking...
21:40 sovetskiy do you here link to some article explaining it in details?
21:41 sovetskiy s/here/have
21:44 guru joined #minetest
21:44 Krock uh well.. classic stackoverflow answer: https://stackoverflow.com/questions/352497​0/why-does-lua-have-no-continue-statement
21:45 deltasquared orbea: well if you're the server owner, you mentioned the config setting, what's up with just using that
21:46 orbea deltasquared: there isn't one, there is no recipe in the mod begin with
21:46 orbea hence why I was asking it to be added
21:46 orbea *mods
21:47 deltasquared orbea: no need to fork, just make a mod which depends on it and define your own crafting recipe
21:47 orbea actually good point...
21:47 deltasquared I've done something similar with a local mod in my singleplayer world to adjust the difficulty.
21:47 orbea thanks
21:48 deltasquared namely, making travelnet elevators require the use of stainless steel, forcing me to advance somewhat in technic before I get easy teleportation
21:48 orbea kind of silly this is needed, but w/e :)
21:48 deltasquared *shrug*
21:48 deltasquared can't speak for that
21:48 orbea yea, it is what it is.
21:49 deltasquared right, must dash, rude interruption time
21:51 sovetskiy unconvenient. what if I want to really break out that loop? I need double break with temporary flag signaling it
21:56 RedSoxFan07 joined #minetest
22:03 RedSoxFan07 joined #minetest
22:08 sovetskiy what global table or method can get me info about fuels? its names? burntimes?
22:11 Krock well, there's minetest.get_all_craft_recipes(query item)
22:13 Krock https://github.com/minetest/minetes​t/blob/master/doc/lua_api.txt#L2932  for reference
22:15 troller joined #minetest
22:21 sovetskiy no
22:21 sovetskiy {
22:21 sovetskiy {
22:21 sovetskiy width = 1,
22:21 sovetskiy type = "normal",
22:21 sovetskiy items = {
22:21 sovetskiy "default:coalblock"
22:22 sovetskiy },
22:22 sovetskiy output = "default:coal_lump 9",
22:22 sovetskiy method = "normal"
22:22 sovetskiy }
22:22 sovetskiy }
22:22 sovetskiy it is for  minetest.get_all_craft_rec​ipes("default:coal_lump")
22:22 sovetskiy there is nothing about its burntime
22:25 Jousway joined #minetest
22:32 Jousway joined #minetest
23:07 EvergreenTree joined #minetest
23:10 proller joined #minetest
23:13 Wuzzy sovetskiy: sounds like a bug
23:32 swift110 joined #minetest
23:34 grumble joined #minetest
23:34 Loquendero joined #minetest

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