Minetest logo

IRC log for #minetest-dev, 2015-08-03

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

All times shown according to UTC.

Time Nick Message
00:00 VanessaE paramat: add savanna biomes to mgv6 :)
00:00 VanessaE (the snow biomes are quite nice, btw)
00:09 zat joined #minetest-dev
00:35 paramat hm kilbith wondered if savanna would be added to mgv6 .. i'm undecided
00:35 hmmmm @param
00:36 VanessaE well now the decision is made.  ad it. ;)
00:36 VanessaE add*
00:36 hmmmm rubywarden noted earlier that the output of mgv6 changed sometime
00:36 paramat yeah i saw that
00:36 hmmmm i'm guessing it's the cave thing
00:36 paramat i tested his seed, it was a small pseudorandom cave
00:36 hmmmm normally i wouldn't care about caves, but they're critical to V6's output
00:36 hmmmm since they shape the terrain so much
00:37 paramat yes
00:37 hmmmm if you change any code in cavegen v6 you need to ensure that the same amount of calls to pseudorandom::next() are being made
00:38 paramat i made a change recently to small caves, will check it
00:39 hmmmm so if you had a conditional statement that you removed, such as if (foobar) { blah = pr.range(0, 5); ... }  ->  blah = pr.range(0, 5); for example, that would be a dangerous modification
00:39 hmmmm it's very delicate
00:40 paramat https://github.com/minetest/minetest/commit/4046f3e302a3394bf376caf543cb643e1562bc5e
00:41 paramat might reverse the 'no small caves entirely above ground' thing
00:41 hmmmm why's that
00:41 paramat as it's not essential, although i can't see how it would add a new cave to ruben's world
00:42 hmmmm like i said, it changes the randomness
00:42 hmmmm so there's a large cave now at that exact position that wasn't there before
00:42 paramat small cave
00:43 hmmmm are you sure it was a small cave?
00:43 paramat yeah so although i didn't expect that commit to change something maybe it did
00:43 paramat yeah sure
00:43 hmmmm https://github.com/minetest/minetest/commit/4046f3e302a3394bf376caf543cb643e1562bc5e#diff-e876a0a129495f2b2dd1182861fa6ea3R444
00:43 hmmmm that return;
00:43 paramat depends on if i understood rubenwardy's description though
00:44 hmmmm the ps2->range() call below is missing, along with all the ps->range() calls in carveRoute()
00:44 hmmmm which is a lot
00:44 hmmmm it's your call though, do you think that v6 cavegen output is okay changed?
00:45 hmmmm the whole point of the "don't change v6 output" is to prevent terrain discontinuities within the same world
00:45 hmmmm but caves are discrete structures
00:45 hmmmm so really the only negative effect would be that the property of having the same exact world given a certain seed is lost
00:45 paramat preserving the terrain is more important than this commit which is non-essential
00:46 paramat so i will reverse that change, and test rubenwardy's seed again to see if there is a change back
00:48 paramat ah of course, a different number of calls to pseudorandom:next, thanks
00:55 paramat that part of the commit only reduces processing load so is non essential
00:56 hmmmm maybe something like bool do_dry_run = p1.Y > h1 && p2.Y > h2;
00:56 hmmmm then in carveRoute()
00:57 hmmmm if (dry_run || !ndef->get(c).is_ground_content) continue;
00:57 leat joined #minetest-dev
00:58 paramat sure maybe later, for now i must quickly fix the terrain
00:58 hmmmm lol it's a 3 line modification
01:04 paramat okay will study the method
01:10 TBC_x are you breaking mapgen after feature freeze?
01:10 TBC_x ;)
01:11 paramat heh luckily i broke it before freeze
01:11 TBC_x you lucky one
01:12 TBC_x gcc compilation takes some time
01:15 TBC_x I should have let it compile just C and C++
01:30 paramat now pushing 2992
01:37 paramat complete
01:44 paramat left #minetest-dev
01:47 leat joined #minetest-dev
01:50 sloantothebone joined #minetest-dev
01:55 sloantothebone How do I make a timer in the form of a variable that I can use an if not statement on?
02:16 Princessheada joined #minetest-dev
02:25 paramat joined #minetest-dev
02:30 zat joined #minetest-dev
02:33 paramat hmmmm, okay i understand now, this means large caves were also altered when i fixed the shadow bugs soon after 0.4.12 stable was released. i will use your 'dry run' method for all caves, for consistency with 0.4.12 stable
03:04 zat joined #minetest-dev
03:49 paramat #3000 hmmmm . i placed the 'continue' as early as possible after the final pseudorandom calls, to minimise processing
03:49 ShadowBot https://github.com/minetest/minetest/issues/3000 -- Cavegen V6: Make all caves consistent with 0.4.12 stable by paramat
03:50 paramat now i need to check if it's similar to 0.4.12 =s
04:00 paramat good grief i wasn't expecting it to actually work, apart from minor expected differences the significant caves are identical to 0.4.12
04:22 paramat will push later if approved
04:22 paramat left #minetest-dev
04:24 lag01 joined #minetest-dev
04:27 hmmmm looks good, approved
05:10 hmmmm heh
05:10 hmmmm lua and luajit have very different ways of handling OOM conditions
05:12 hmmmm if I get this working it looks like I'm gonna have to make vanessae downgrade to interpreted lua :I
05:13 hmmmm the great part is that i think i can reliably handle out-of-memory in a very controlled fashion, it's just that under linux the oom killer needs to be disabled somehow
05:35 paramat joined #minetest-dev
05:35 paramat now pushing 3000
05:43 paramat complete
06:06 Hunterz joined #minetest-dev
06:33 GunshipPenguin joined #minetest-dev
06:43 paramat game#606
06:43 ShadowBot https://github.com/minetest/minetest_game/issues/606 -- Flowers: Make mushrooms farmable using spores WIP by paramat
06:56 Puma_rc joined #minetest-dev
07:00 hmmmm okay nevermind
07:00 hmmmm i can do this with luajit but it needs to be modified
07:04 Darcidride joined #minetest-dev
07:11 Puma_rc joined #minetest-dev
07:16 Puma_rc joined #minetest-dev
07:21 paramat left #minetest-dev
07:52 nrzkt joined #minetest-dev
08:00 Yepoleb_ joined #minetest-dev
08:03 kilbith joined #minetest-dev
08:18 Calinou joined #minetest-dev
08:43 SopaXT joined #minetest-dev
08:45 SopaXT joined #minetest-dev
08:47 Amaz joined #minetest-dev
09:29 rubenwardy joined #minetest-dev
09:32 H-H-H joined #minetest-dev
09:34 kilbith ~seen RealBadAngel
09:34 ShadowBot kilbith: I saw RealBadAngel in #minetest-dev 1 week, 2 days, 2 hours, 1 minute, and 27 seconds ago saying "to fit android devices"
09:35 nrzkt ~seen NullPointerException
09:35 ShadowBot nrzkt: Error: 'NullPointerException' is not a valid nick.  That nick is too long for this server.
09:35 nrzkt ok this is good :D
09:35 nrzkt IRC refuses NullPointerExceptions :D
09:36 kilbith seen != grep
09:40 leat joined #minetest-dev
09:40 lag01 joined #minetest-dev
09:55 bluegreen joined #minetest-dev
09:57 rubenwardy 2015-08-03 10:55:52: ERROR[main]: generateImage(): Could not load image "disable_img.png" while building texture
09:57 rubenwardy rubenwardy@rubenPC:~/dev/minetest/src$ grep -nr "disable_img" *
09:57 rubenwardy Binary file CMakeFiles/minetest.dir/mapblock_mesh.cpp.o matches
09:59 VanessaE that's one of RBA's shader flags
10:00 VanessaE (where he needs a bool from the main code, he passes an all-white or all-black image to the shader)
10:06 rubenwardy it should give that chat error message, though
10:06 rubenwardy *shouldn't
10:07 msantana` joined #minetest-dev
10:14 TBC_x hmm... How do I pass alternative compiler path to cmake?
10:16 VanessaE -DCMAKE_INSTALL_PREFIX=/usr
10:16 TBC_x that's not what I'm looking for
10:18 VanessaE oh, thought you meant the install path
10:19 TBC_x I had to compile gcc on my mint machine
10:20 VanessaE no idea then :)
10:22 nrzkt poudriere bulk -p default -j jailname lang/gcc
10:22 nrzkt oops you are on a linux :D
10:22 TBC_x I spent six hours last night putting the machine back into bootable state
10:22 nrzkt if you want to compile, use gentoo
10:23 diemartin joined #minetest-dev
10:25 Darcidride_ joined #minetest-dev
10:38 TBC_x Do server really need irrlicht?
10:41 nrzkt it's because of v3f and other irrlicht related defines
10:43 nrzkt who reviews https://github.com/minetest/minetest/commit/79e2647556773d46cc0299e73ca23c41a153ff2a ?
10:43 nrzkt a mutex on getPeerIDS, why not. But no mutex in setPeerID and getPeerId ?
10:45 nrzkt okay, if two thread wants to read simultaneously we block, but we let two thread to read and write the value together...
10:45 nrzkt this commit is just... useless
10:47 TBC_x in connection.cpp?
10:47 H-H-H connection.h
10:48 TBC_x it needs to be rewritten
10:48 TBC_x from scratch
10:48 nrzkt i agree with you there are problem on it
10:49 nrzkt but the commit done has no effect
10:50 Darcidride joined #minetest-dev
10:55 TBC_x hmm... what a standalone built server needs to get it running?
10:55 Wayward_One joined #minetest-dev
11:09 H-H-H joined #minetest-dev
11:09 TBC_x VanessaE: how does your dreambuilder game work with mapgenv7?
11:09 VanessaE I've never tried it
11:09 TBC_x ok
11:09 VanessaE but as I understand, it generally works ok
11:09 VanessaE since the node names are the same, it should be fine (plants_lib and related mods are the only real concern)
11:10 VanessaE glooptest and a couple other mods in the game need rewritten to use the modern ore system but that's about it.
11:11 TBC_x I just need some crazy mods to crash the server so I can extract logs
11:11 VanessaE give it a shot then :)
11:12 VanessaE try downloading my "Survival" server map.  if anything will stress a server, that will.
11:13 H-H-H just out of curiosity how many nodes can you have inworld before it says stuff you ive had enough and dies
11:14 TBC_x using btrfs was a mistake
11:15 * H-H-H is re trying master for android to see if est31 has fixed the bugs i showed him
11:16 TBC_x fatal: unable to overwrite old ref-pack file: No space left on device -_-
11:17 VanessaE eep
11:18 VanessaE H-H-H: the internal hard node defs limit is 65534.  It's soft-limited to 32766.
11:18 VanessaE Dreambuilder uses a little over half of that soft limit
11:19 H-H-H ty VanessaE
11:21 VanessaE the practical limit on the other hand is up to your machine's RAM amount
11:21 leat joined #minetest-dev
11:21 VanessaE Dreambuilder needs about 1.6GB or so for its ~17k node defs and textures and so on.
11:23 H-H-H intersting that there is a hard limit, i would have thought that with the right hardware it "could" be infinate
11:24 VanessaE it can be, but the variables related to the node count are unsigned 16-bit.
11:24 TBC_x do we even support node variations?
11:24 H-H-H ahhhhhhhhhh yeah that makes sence
11:24 VanessaE TBC_x: only "special tiles".
11:24 VanessaE and then that's extremely limited
11:24 VanessaE so essentially, no.
11:25 VanessaE I'm not sure about the technical underpinnings of the 16-bit figure, but were it raised to, say, 32-bit unsigned, you could say it's effectively unlimited.
11:26 VanessaE TBC_x: there's an issue calling for varied textures via a sprite strip image similar to how animations work, and RBA wants to get something pushed through to allow alternate node defs to be called upon by changing a bit in param2 (I think).
11:27 VanessaE and there was that meta_set_nodedef branch, which would basically decouple an in-map node from its default nodedef, so you could say set_node() cobble stairs, and then meta-them into a straw slab without touching the cobble stairs node def.
11:27 VanessaE but that one never went in either - it wasn't considered finished, but it DID work.  nore made a test mod to use it.
11:28 err404 joined #minetest-dev
11:29 TBC_x when minetest can run full blown RedPower2 I'd consider it finished :)
11:30 * H-H-H waits for hell to freeze over first
11:31 VanessaE TBC_x: well, there IS the "wires everywhere" mod.. does that help? :)
11:31 leat joined #minetest-dev
11:32 VanessaE Here's that test mod I mentioned:  https://github.com/Ekdohibs/microblocks   and here's that wires mod:  https://github.com/Ekdohibs/wires
11:33 TBC_x I mean features like wires and pipes in the same node
11:33 VanessaE (the meat of the test mod is here: https://github.com/Ekdohibs/microblocks/blob/master/init.lua#L71 )
11:33 VanessaE meta_set_nodedef could theoretically do that.
11:34 TBC_x awesome
11:35 TBC_x It did not run out of space this time!
11:36 TBC_x can any of the mods escape out of minetest sandbox? like mess with the filesystem?
11:37 VanessaE not if you have that mod security code activated
11:37 VanessaE otherwise yes.
11:37 TBC_x will create separate user for mt
11:37 TBC_x that should do it
11:37 VanessaE yep
11:38 VanessaE none of DB's mods try to do that, but nevertheless, I run my servers as user "minetest" with appropriate directory ownership here and there.
11:39 TBC_x I have a script somewhere that used to set up minecraft server user and all proper groups and masks so anyone with the gid could mess with it
11:40 TBC_x hmm... can I redirect all the media downloads to your server? and how? :)
11:40 TBC_x 256kb upload is not gonna work with me
11:41 VanessaE go ahead
11:42 VanessaE just point your remote_media at http://minetest.digitalaudioconcepts.com/creative-survival-media/  or  http://minetest.digitalaudioconcepts.com/basic-media/
11:42 VanessaE (the former is dreambuilder, the latter is minetest_game, each plus a handful of mods)
11:43 H-H-H hmmm i wonder if est31 tested this android build
11:43 VanessaE isn't that nrzkt's area?
11:43 H-H-H both from what i understand
11:44 H-H-H i just built current head for android and it doesnt even start lol the last time i built it it ran but had a big problem with sound
11:44 TBC_x duh... I have to prefix mt binary with  LD_LIBRARY_PATH=/usr/local/lib64
11:45 TBC_x otherwise it cannot load libasan
11:46 TBC_x any example minetest.conf I can hold on?
11:46 TBC_x that is also usable
11:47 nrzkt VanessaE: i only build and distribute on playstore.
11:47 VanessaE ok
11:48 H-H-H est31 does fdroid i belive
11:48 nrzkt yes
11:48 H-H-H ok who uninstalled eclipse lol now i cant even use adb
11:51 nrzkt i didn't uninstall eclipse atm to use android studio, need to migrate projects. and adb is included in many distros if you install android-sdk instead of using eclipse installation.
11:51 nrzkt yaourt -S android-sdk on arch
12:10 * H-H-H was j/k i already have the sdk or i wouldnt have been able to build it :P
12:23 TBC_x wow
12:24 TBC_x I've got 100% unhelpful error message
12:24 VanessaE "you don't say?"  :O
12:27 TBC_x what can cause it to `terminate called after throwing an instance of 'ServerError'`
12:27 VanessaE -that's usually a mod error.
12:27 VanessaE most of the time anyway.
12:27 TBC_x with what():  Failed to initialize world
12:27 TBC_x how to make the serve to create the world?
12:28 VanessaE ok THAT is probably not a mod error :P
12:28 VanessaE out of space again?
12:28 TBC_x hopefully not
12:29 VanessaE I don't know off the top of my head, so what I did originally was to create a world in my client, delete the map and ancillary files created by mods, and save the result as a "blank template"
12:31 VanessaE then I just copy it, modify the seed and mapgen params as needed, and use that as the new world :P
12:32 TBC_x creating a server should require minimal effort
12:32 TBC_x creating a basic server should require minimal effort
12:43 alket joined #minetest-dev
13:03 TBC_x finally
13:08 TBC_x whoops
13:08 TBC_x I've got a world made of undefined nodes
13:10 bluegreen How did you do that?
13:11 TBC_x I copied random world folder from my .minetest/worlds
13:11 TBC_x that world was probably mt 3.x
13:12 blaise joined #minetest-dev
13:12 TBC_x http://minetest.digitalaudioconcepts.com/basic-media/castle_ironbound_chest_front.png not found (HTTP response code said error) (response code 404)
13:12 TBC_x is that path ok?
13:13 TBC_x VanessaE: ^^
13:16 TBC_x src/mapgen_v6.h:171 leaks
13:18 kilbith leaks everywhere, we should have strainer as logo
13:18 TBC_x haha
13:18 TBC_x I fixed a bunch of them
13:19 kilbith and RBA leaving before the release, pfff
13:20 TBC_x I think we should disable minimap radar mode when it is broken
13:20 TBC_x or fix it
13:20 TBC_x doxygen is very useful while tracking leaks
13:21 TBC_x and also, I should send memory leak patch to fluidsynth guys
13:21 kilbith what about a garbage collector for C++ ?
13:21 kilbith no longer need to waste time in investigation
13:22 TBC_x ruins determinism
13:22 TBC_x and gc is not silver bullet
13:22 kilbith https://en.wikipedia.org/wiki/Boehm_garbage_collector
13:23 TBC_x RAII should be the GC
13:23 TBC_x RAII is the only thing I would use C++ instead of C
13:28 VanessaE TBC_x: any time your client reaches for a file that's not on the remote media server, the cURL part will throw an error and the client will fetch it directly from the minetest server instance instead.
13:30 TBC_x oh
13:46 TBC_x what setting disables item preloading?
13:47 TBC_x because all that stuff cannot fit into 2GiB of RAM
13:49 Megaf joined #minetest-dev
13:49 TBC_x oh nvm
13:54 RealBadAngel joined #minetest-dev
13:54 RealBadAngel hi
13:54 kilbith ah, he's finally here !
13:55 nrzkt hi RealBadAngel
13:55 RealBadAngel sorry, had some hard days irl
13:55 nrzkt no problem, each have an IRL (i hope)
13:56 RealBadAngel i was reading logs so i know whats going on
13:56 kilbith TBC_x has fixed some memleaks
13:57 RealBadAngel definitely i need to learn that valgrind thingy
13:57 OldCoder joined #minetest-dev
13:58 kilbith and hmmmm says you can improve the minimap algorythm
13:58 TBC_x how do I set a spawnpoint in config?
13:58 RealBadAngel yes, im working on it but its long term work, since im rebuilding minimap core
13:58 RealBadAngel including scanning
13:59 kilbith TBC_x: https://github.com/minetest/minetest/blob/master/minetest.conf.example#L377
13:59 RealBadAngel so no chance for 0.4.13
13:59 TBC_x I really don't want to open a web browser right now
13:59 kilbith static_spawnpoint = 0, 0, 0
13:59 TBC_x thanks
14:00 RealBadAngel for 0.4.13 i would like to fix a few things, like fsaa + plants and some minor glitches with shaders
14:01 RealBadAngel also imho mt game shall get complete displacement maps before release
14:03 TBC_x what units the static_spawnpoint accepts? BS?
14:06 RealBadAngel btw, i was testing lately my new tablet
14:06 RealBadAngel controls there are horribl
14:06 RealBadAngel e
14:07 RealBadAngel for touchscreen there should be a "joystick" field instead of buttons
14:08 RealBadAngel ie when you touch the field moving a finger will cause movement into that direction, released - back to center position
14:08 nrzkt RealBadAngel: agree
14:10 RealBadAngel now i do have quite good tablet so i can work on it
14:10 RealBadAngel minetest is working quite nicely on it - galaxy tab 4 10''
14:11 TBC_x looks like we are leaking initialized nodes
14:11 RealBadAngel what do you mean?
14:11 TBC_x whatever happens behind the scenes when the loading screens says Initializing nodes...
14:12 TBC_x hmm
14:13 RealBadAngel theres lotsa weird things going on during that time
14:13 TBC_x I should mention that I'm using 4.12 release client for this
14:14 RealBadAngel when it comes to gfx 0.4.12 is hardly compatible with current master
14:14 RealBadAngel theres lotsa things added or changed
14:20 VanessaE a wile RBA appears!
14:20 VanessaE wld*
14:20 nrzkt gotcha !
14:22 RealBadAngel VanessaE, btw, who has banned me on your server? :P
14:22 VanessaE RealBadAngel: I wasn't aware that anyone had.
14:22 RealBadAngel i was called imposter and kicked out of it :P
14:22 VanessaE oh, I'll fix it.  which one?
14:22 TBC_x 90.176.17.32:30000
14:22 VanessaE hm, I see no ban record for you on VE-Basic.
14:23 RealBadAngel it was like 2 days ago, was using nick 1RealBadAngel
14:23 VanessaE moment..
14:23 VanessaE ah there it is.
14:23 VanessaE fixed.
14:39 bluegreen joined #minetest-dev
14:39 Gael-de-Sailly joined #minetest-dev
14:46 crazyR joined #minetest-dev
14:46 crazyR joined #minetest-dev
14:58 blaise joined #minetest-dev
15:15 zat joined #minetest-dev
15:32 zat joined #minetest-dev
15:38 chainsaw10 joined #minetest-dev
16:00 TBC_x god damn SharedBuffer
16:01 TBC_x heap-use-after-free
16:04 TBC_x http://sprunge.us/GEdS
16:06 TBC_x interesting is, that I never encountered this on the client
16:08 hmmmm joined #minetest-dev
16:22 Hunterz joined #minetest-dev
16:31 VanessaE http://pastebin.com/3sGMzmHy
16:31 VanessaE TBC_x: you mean like that?
16:31 VanessaE it happened just now
16:37 luizrpgluiz joined #minetest-dev
16:38 luizrpgluiz what is the probability that the minetest has the support ncurses?
16:42 sloantothebone joined #minetest-dev
16:42 RealBadAngel joined #minetest-dev
16:42 luizrpgluiz or commands within the minetest server, without being within the game, the server terminal? I do not want to have to install Linux only porcausa the mod irc
16:42 sloantothebone joined #minetest-dev
16:44 luizrpgluiz porcausa = just because ^^
16:44 TBC_x luizrpgluiz: someone is workin on that
16:50 luizrpgluiz this is very good because it has servers and users who do not want to install another operating system just because the irc mod only works on it
16:52 TBC_x what?
16:54 VanessaE TBC_x: [08-03 12:00] <TBC_x> god damn SharedBuffer  <-- anything like this?  http://pastebin.com/3sGMzmHy
16:54 TBC_x yes
16:55 VanessaE ok
16:55 VanessaE that crash happened on VE-Basic about 25 mins ago
16:56 TBC_x luizrpgluiz: what?
16:58 luizrpgluiz IRC mod uses the Linux operating system to work and has many people who do not like Linux to use mod
16:58 VanessaE TBC_x: what he means is that the only way to get "console" access to a minetest server is via the IRC and irc_commands mods, which don't work on Windows.  As he's stuck with that OS, he wants the previously-discussed ncurses console.
16:58 TBC_x oh
17:08 luizrpgluiz yes Vanessa
17:09 luizrpgluiz ncurses is well developed, could help many server administrators
17:11 est31 joined #minetest-dev
17:15 TBC_x the guy who implemented SharedBuffer<T> thought how much is MT single-threaded
17:17 est31 H-H-H, if your android build fails, you will perhaps have to remove your deps/ directory, or do a clean re-build
17:18 est31 the commit that fixed the sound has changed the ABI, so the linker would refuse linking.
17:31 lag01 joined #minetest-dev
17:43 H-H-H est31 it builds fine just doesnt run
17:44 est31 did you do a clean rebuild?
17:44 est31 also whats the adb error?
17:52 nrzkt joined #minetest-dev
18:03 TBC_x what is the point of  ReliablePacketBuffer.m_oldest_non_answered_ack ?
18:04 TBC_x isn't that supposed to be always the last packet in the list?
18:06 sloantothebone joined #minetest-dev
18:06 TBC_x it looks like write-only member
18:07 H-H-H sorry yes i did a completely clean clone and build and adb is being a pia atmo and wont even list my device lool
18:09 hmmmm TBC_x:  I realize that connection.cpp could do better with regards to thread safety, but I legitimatly cannot find another thread that accesses outgoing_reliabes_sent other than ConnectionSendThread
18:09 hmmmm maybe I'm mistaken and I missed something
18:09 sloantothebone joined #minetest-dev
18:10 sloantothebone joined #minetest-dev
18:11 kahrl TBC_x: I guess m_oldest_non_answered_ack was used previously but some refactoring removed the reads
18:12 TBC_x I'm completely removing it
18:12 TBC_x causing confusion
18:13 MinetestForFun joined #minetest-dev
18:15 hmmmm well I don't have any theories about how this happens
18:15 hmmmm a race condition is an easy target, but it logically can't happen here
18:15 kahrl heh, looks like it was already unused in the commit that added it
18:15 hmmmm maybe it would help if we kept a record of each location a reference was grabbed
18:16 TBC_x it is afaik only happenging with BufferedPacket
18:17 hmmmm but honestly, i'm not sure i understand why BufferedPacket needs to have a SharedBuffer at all
18:17 hmmmm it's not shared with anything
18:17 kahrl https://github.com/minetest/minetest/commit/9edb91da5754cf194637d1d7faa513719b61f9b4
18:17 TBC_x yes
18:17 TBC_x I've removed it in my code
18:17 TBC_x haven't run it yet though
18:18 hmmmm ?
18:18 hmmmm you mean removed the Shared part of the SharedBuffer?
18:18 TBC_x used simple Buffer<T>
18:18 hmmmm yeah
18:18 hmmmm sounds reasonable
18:18 hmmmm instead of figuring out what's accessing it in a weird manner, though, it'll just make another copy :/
18:18 hmmmm but it won't crash anymore at least!
18:19 hmmmm meh who cares, connection.cpp is the epitome of inefficiency
18:19 TBC_x you don't say
18:20 H-H-H est31 http://pastebin.com/cD7nK7jK
18:20 TBC_x JMutex is imho too limited to do anything efficient
18:20 H-H-H finally got adb to play nicely
18:20 hmmmm it's not JMutex that's inefficient
18:20 hmmmm it's all the copies of things
18:20 hmmmm look at how often Buffers are shared around without references
18:21 hmmmm receiving a packet iirc does somewhere around 13-15 copies of the same exact data
18:22 TBC_x given how clear is the architecture of the code there, I'm not surprised
18:22 est31 H-H-H, seems you have iconv problems
18:22 H-H-H looks to be libiconv.so not behaving
18:22 est31 H-H-H, do you have a log of the build?
18:22 hmmmm GERMAN ENGINEERING
18:23 H-H-H i can prob get the libiconv portion from me bash scrollback or i can do a rebuild and redirect it to a file
18:23 est31 make sure you make &> not just Y
18:23 est31 err
18:23 est31 >
18:24 est31 <> and Y are adjacent on german keyboards :)
18:24 H-H-H ok will do a clean rebuild and get back to you :)
18:24 bluegreen joined #minetest-dev
18:25 hmmmm TBC_x:  Considering the fact that connection.cpp is in dire need of a rewrite I'd be willing to accept changing SharedBuffer to Buffer as a solution
18:28 est31 hmmmm, french or polish ain't better :p
18:28 hmmmm sapier was the one who did connection.cpp
18:28 hmmmm heh
18:28 hmmmm it's overengineered
18:28 hmmmm too complicated and kinda scattered
18:29 TBC_x when I'll see sapier, I'll act like his code
18:29 TBC_x throw something at him
18:29 H-H-H ok build underway
18:29 H-H-H completely deleted and recloned
18:30 est31 nice
18:30 hmmmm he's extremely prolific
18:30 hmmmm sapier did the entire lua mainmenu api in about 3 days
18:30 est31 nrzkt, what happened with the android buildbot?
18:30 est31 nrzkt, can you reactivate it?
18:32 nrzkt oh i disabled it because of sourceforge
18:32 nrzkt every build failed because of download fail
18:33 TBC_x the SharedBuffer could also fail because of try ... catch blocks everywhere
18:34 H-H-H why not just change the makefile to get irrlicht from github
18:35 est31 it works now again
18:35 est31 github can have outages too
18:35 hmmmm are you sure about that?  when an exception is caught, the SharedBuffer falls out of scope, decreasing its refcount
18:35 est31 e.g. some chinese evil guys changing ad scripts
18:35 est31 to test their new cyberweapon
18:35 TBC_x I'm not sure how well it is handled 'unmutexed'
18:36 H-H-H yeah true est31 but github appears to be a lot more reliable than sf
18:36 est31 H-H-H, but thats the official source
18:36 TBC_x with the connection.cpp in such state, I'm not sure about anything
18:36 est31 all others are just inofficial clones
18:36 H-H-H bah lol build failed on d/l freetype lol the remote end hung up unexpectedly so re running build
18:37 H-H-H thats better its doing its thing now
18:37 nrzkt okay est31: re-enable build then
18:37 est31 VanessaE, you can see the proposed algorithm described in #3003
18:37 ShadowBot https://github.com/minetest/minetest/issues/3003 -- Do leafdecay scanning in the engine
18:37 est31 I hope its understandable how the speedup is gained
18:38 nrzkt est31: it's re-enabled now
18:39 luizrpgluiz left #minetest-dev
18:43 TBC_x 3003 seems to me like a hack
18:44 est31 what would you propose?
18:45 TBC_x replace abm
18:45 est31 by?
18:46 TBC_x behaviour patching and timers
18:46 est31 please explain
18:48 TBC_x for example, water flow, It is a behaviour of water node dictating where to flow depending on some conditions. When a mod adds for example a sponge, it patches water flow behaviour to not flow into its defined area.
18:49 est31 now tell me the benefit
18:49 est31 and how to implement it
18:50 est31 also it has nothing to do with 3003
18:50 est31 3003 is about improving an existing mod api method
18:51 est31 your proposal is about adding a new one
18:52 est31 code is behaviour, so abms are precisely that
18:52 est31 with an abm you define behaviour
18:52 est31 in an abstract way
18:52 TBC_x you know what?
18:52 TBC_x I should read first, then talk
18:55 TBC_x So if I understand that right, you propose to notify all neighbours of a node, whenever the node itself gets somehow invalidated?
18:56 est31 its about making leafdecay fast
18:56 est31 e.g. you have a tree, and remove the trunk nodes
18:57 est31 then the leaves fall down
18:57 est31 to make that fast
18:58 TBC_x I think that range of 1 could suffice
18:59 est31 nope, trees can be huge
18:59 est31 seen moretrees trees?
19:00 TBC_x are currently the leaves scanning their surroundings each tick?
19:00 est31 yes
19:00 TBC_x I think this could be trigger-based
19:01 est31 to make it bearable, the moretrees leaves scan is made very slow.
19:01 TBC_x and possibly make nodes to communicate with each other
19:01 est31 TBC_x, agreed, the cache is only invalidated if there is a change.
19:02 crazyR joined #minetest-dev
19:05 alket joined #minetest-dev
19:07 TBC_x hmm... Instead of scanning, the nodes theoretically could channel the invalidation event with set amount of hops
19:08 TBC_x but I'm not sure if this woul've been useful
19:09 est31 invalidation should be fast, so that you can edit multiple nodes, before an abm step happens
19:10 est31 if it involves scanning about what precisely to invalidate, it perhaps spares time on the re-validation, but it also creates more overhead at the "editing" part.
19:11 est31 also, I guess we won't need complete invalidation then, we could just edit the cache.
19:15 sloantothebone How do I detect the presence of a mod in an if statement?
19:16 est31 there is no unified way
19:17 est31 there has been some "add warnings" commits, but the proposed way to do it has flaws too.
19:20 Puma_rc joined #minetest-dev
19:22 sloantothebone Ok, I want this code to activate only if the waterplus mod is activated: http://pastebin.com/WKWGXb4A
19:24 sloantothebone Ok... All I need is "if minetest.get_modpath("waterplus") then" but where do I put it?
19:24 est31 around the register_abm
19:27 TBC_x there is so much code depending on connection.h
19:29 alex_m joined #minetest-dev
19:32 alket_ joined #minetest-dev
19:33 sloantothebone Ok I did
19:36 sloantothebone Ok can I have a quick tutorial on how to post this on github?
19:37 sloantothebone Do I have to log in with the git command?
19:37 est31 yes
19:38 est31 https://help.github.com/articles/fork-a-repo/+
19:38 est31 err
19:38 est31 https://help.github.com/articles/fork-a-repo
19:38 est31 and https://help.github.com/articles/good-resources-for-learning-git-and-github/
19:39 est31 hmmmm, there was a patch in 2006 I think which allows you to edit some file in procfs to disable the oom killer
19:40 H-H-H est31 https://www.dropbox.com/s/k67ialor1qjba7w/build.txt?dl=0 complete build log
19:43 sloantothebone Ooh I have rabbitvcs git options in right click
19:47 crazyR joined #minetest-dev
19:47 alket im getting this shadow when using WorldEdit http://i.imgur.com/11DkPGD.jpg
19:51 H-H-H any clues est31 ?
19:53 est31 H-H-H, moment pls I'm translating a game for Calinou.
19:53 paramat joined #minetest-dev
19:53 H-H-H ok np :)
19:54 crazyR_ joined #minetest-dev
19:57 paramat hi sfan5 any comments on game#604 game#605 game#606 ?
19:58 ShadowBot https://github.com/minetest/minetest_game/issues/604 -- Stairs: Lengthen interval of replace abm by paramat
19:58 ShadowBot https://github.com/minetest/minetest_game/issues/605 -- Default/functions: Reduce lavacooling ABM/sound overload by paramat
19:58 ShadowBot https://github.com/minetest/minetest_game/issues/606 -- Flowers: Make mushrooms farmable using spores WIP by paramat
19:58 sfan5 oh right
19:58 sfan5 you weren't there earlier
19:58 sfan5 paramat: 605 is ok, 604 with the new interval you suggested
19:58 sfan5 looking at 606 now
19:59 luizrpgluiz joined #minetest-dev
20:00 sfan5 hm
20:00 sfan5 flower spores?
20:00 sfan5 I'm not sure
20:00 luizrpgluiz as I do I read .mts files in text mode?
20:01 luizrpgluiz because I wanted to modify the tree schematics
20:02 alket what about my error :)
20:03 sfan5 luizrpgluiz: .mts files are not text files an compressed
20:03 sfan5 and*
20:04 luizrpgluiz but as I do I read them?
20:07 kahrl alket: questions about specific mods go to #minetest
20:07 alket kahrl: thanks
20:07 sloantothebone How do I add a depends.txt with git?
20:08 H-H-H touch depends.txt
20:08 H-H-H git add .
20:08 H-H-H git commit -m"comitt msg here"
20:08 H-H-H git push
20:11 sloantothebone Ok why didnt that work when I rightclicked and clicked commit, because I have rabbitvcs git options in nautilus
20:12 sloantothebone Wait a minute it still hasn't changed on github
20:13 sloantothebone https://github.com/sloantothebone/minetest-mod-weather/blob/master/weather/rain.lua
20:13 H-H-H did you push it
20:13 sloantothebone still no depends.txt
20:13 H-H-H did you push it
20:13 paramat okay thanks
20:13 est31 H-H-H, can you paste the content of build/android/deps/libiconv/lib/Makefile ??
20:13 sloantothebone Ok pushing it
20:13 sloantothebone Oh forgot the difference between push and pull
20:14 paramat sfan5 the spores are for mushrooms only
20:15 H-H-H est31 http://pastebin.com/sA1BqLGe
20:15 sfan5 only mushrooms
20:15 sfan5 oh
20:15 * paramat reads logs
20:16 sfan5 paramat: ok if you do what you suggested in your last issue comment
20:16 paramat okay, i guess '..' needs spaces
20:20 est31 H-H-H, whats the output of readelf -a build/android/libs/armeabi-v7a/libiconv.so | grep SONAME
20:22 H-H-H 0x0000000e (SONAME)                     Library soname: [libiconv.so]
20:22 est31 thats ok then
20:22 est31 hrmmmm
20:22 H-H-H maybe a packaging error ?
20:23 sfan5 est31: i have no idea what the problem is but maybe try running ldd on the minetest executable/lib and see what it says
20:24 H-H-H it would seem libiconv.so is missing from a freshly built apk
20:25 est31 So the output of unzip -l build/android/bin/Minetest-debug.apk | grep iconv
20:25 est31 is empty?
20:26 H-H-H sorry est31 that was just my observation which was just proved wrong
20:26 H-H-H 914576  2015-08-03 20:31   lib/armeabi-v7a/libiconv.so
20:30 H-H-H fwiw est31 i also tried a release build and it was the same
20:33 est31 H-H-H, what happens if you add System.loadLibrary("iconv"); to build/android/src/net/minetest/minetest/MtNativeActivity.java
20:34 H-H-H give me a few and i will try it
20:34 est31 e.g. line 90
20:36 crazyR joined #minetest-dev
20:36 kilbith joined #minetest-dev
20:39 H-H-H is a make clean and then make sufficent ?
20:39 est31 not even clean needed
20:40 est31 so yes :
20:40 est31 )
20:40 H-H-H ok testing new apk
20:40 VanessaE est31: your leaf decay idea makes sense, I just hope those proposed arrays don't take up a ton of RAM in practice (imagine 20+ leaf decay routines running, in addition to regular ABMs)
20:41 est31 VanessaE, how much blocks are loaded "in practice" ??
20:41 VanessaE around 1500.
20:41 est31 but yeah, its the old fight, memory vs speed :)
20:41 VanessaE sometimes more
20:42 VanessaE (I've not taken a recent count, that ^^^ is just what I remember from some tests a long while ago)
20:43 est31 and how many abms do run?
20:43 est31 leafdecay?
20:43 VanessaE probably 20 as a rough guess.
20:44 est31 !c 20 * 1500
20:44 ShadowBot est31: math calc <math expression>
20:44 H-H-H ok it runs but i have no txt in mainmenu
20:44 VanessaE there are around 15 tree types, plus the default one, plus whatever extras are added (like farming_plus's banana tree)
20:45 est31 H-H-H, ok, this is then the bug #2973
20:45 ShadowBot https://github.com/minetest/minetest/issues/2973 -- [Android] No text on main menu
20:46 H-H-H also it didnt create Minetest folder but minetest
20:47 est31 did it?
20:47 est31 interesting
20:48 H-H-H yeah but then every version i have built except 3.12-stable has done the same
20:48 H-H-H sorry 4.12-stable
20:49 est31 well, the filesystem we write it to is case insensitive
20:49 est31 but it shouldnt do it
20:51 H-H-H yeah since android is linux skinned lol
20:52 est31 no, we write to fat32
20:55 TBC_x are moretrees sequoias supposed to cause long shadow?
20:55 VanessaE TBC_x: nope.
20:55 VanessaE only a simple vertical shadow like anything else
20:55 VanessaE a fairly dark one, to be sure, but that's all
20:55 TBC_x it is mapgenv6 bug then...
20:58 H-H-H at the risk of sounding stupid i thought android used ext4 with the exception of some samsung devices
20:58 TBC_x well, you can join 90.176.17.32:30000 and take a look at it if you want
20:58 est31 H-H-H, its stored on the sd card
20:58 est31 and thats fat32
20:58 H-H-H mine isnt
20:58 TBC_x damn windows
20:58 est31 in order to work with stupid windows OS devices
20:59 TBC_x windows is blocking innovations
20:59 H-H-H i dont own any windows devices and have always formated my memory cards tyo be used with linux
20:59 est31 can apple mount ext4??
21:02 est31 man this sentence is stupid
21:03 est31 " The EXT file system (short for Extended File System) and it’s family members of EXT2, EXT3, and EXT4, are the file systems used by Linux and Raspberry Pi."
21:03 est31 http://osxdaily.com/2014/03/20/mount-ext-linux-file-system-mac/
21:05 luizrpgluiz left #minetest-dev
21:14 TBC_x damn, mobs in dreambuilder are scary
21:15 VanessaE they're from Calinou's carbone_mobs
21:17 TBC_x wtf is trooper?
21:18 Calinou a monster that looks like a regular player
21:18 TBC_x didn't know I can pick him up
21:20 TBC_x the animations lack interpolation
21:21 Calinou like all Minetest animations...
21:21 Calinou we disabled animation blending due to an Irrlicht bug
21:21 TBC_x well... that's horrible
21:21 est31 VanessaE, it will use around 17 MB
21:22 est31 for 1.5 k loaded blocks
21:22 VanessaE ok
21:22 est31 can be easily reduced though, depends on how the storage is implemented
21:22 VanessaE that's not too bad.  is there anywhere we... yeah
21:22 est31 reduced by factor of 8
21:22 TBC_x but not just animations, at least movement could use some interpolation
21:22 est31 because with that calculation, every boolean takes up one byte
21:22 est31 thats pretty wasty :)
21:23 TBC_x I doubt that is irrlicht bug
21:23 Calinou TBC_x, animation frames are already interpolated
21:23 Calinou but animations don't have interpolation between them
21:23 est31 I wonder whether we even need iconv on android
21:23 est31 I just wonder what irrlicht's magical wchar format is
21:23 est31 then I know which converter to use.
21:24 TBC_x Calinou, do your mobs despawn?
21:24 est31 H-H-H, whats your ndk version??
21:24 Calinou they do, if player is far away enough
21:25 TBC_x duh... I want to keep him
21:25 TBC_x could use some 'undespawnable' tag
21:26 TBC_x milestone for mt 0.4.14 should be smooth everything and connection.cpp rewrite
21:28 H-H-H r9d
21:29 est31 TBC_x, whoever rewrites connection.cpp has to be trusted. this isnt something that should be rushed in any way.
21:29 est31 also, I dont think we should add 0.4.14
21:30 est31 after 0.4.13, we should either make 0.5.0 or 1.0
21:30 est31 and stop doing minor and patch versions
21:30 est31 but rather major and minor
21:31 TBC_x 0.x.x means unstable, unfinished to me
21:32 Calinou we should do like Kernel Adiutor
21:32 Calinou who doesn't want 0.9.4.6.1?
21:35 sloantothebone How do I check if a string is not "false"? "and RAIN_DROPS <> "false" then" or "and not RAIN_DROPS == "false" then"?
21:35 VanessaE sloantothebone: #minetest
21:35 VanessaE (and it's ~= for 'not equals')
21:36 est31 VB.net <> lua <> sql
21:40 est31 H-H-H, can you upload the generated apk somewhere??
21:41 H-H-H i can probably put it on dropbox
21:54 nrzkt est31: i don't think 0.4.13 is a 0.4 milestone due to many changes, unlike 0.4.12 which is a regular 0.4.x series update
21:55 nrzkt 0.5 to mark fixes, minimap adding, authentication strength is a good thing :)
22:03 GunshipPenguin joined #minetest-dev
22:04 diemartin joined #minetest-dev
22:05 Player_2 joined #minetest-dev
22:14 TBC_x damn... I thought I fixed that
22:15 TBC_x http://sprunge.us/YUZU
22:17 crazyR joined #minetest-dev
22:26 TBC_x oh, that's a different one
22:34 bluegreen joined #minetest-dev
22:40 VanessaE TBC_x, hmmmm:  here's another one, happened just now:  http://pastebin.com/k5QFcZQ2
22:44 TBC_x VanessaE: that one should have been fixed by abe6c07
22:45 VanessaE my last restart (due to a crash) was ~6 hours ago, and that brought in a build that's 2 commits newer.
22:45 TBC_x if the crash was caused by Thread 4
22:46 VanessaE it was at 7a6e4dc5 at the moment of that crash
22:46 TBC_x darn
22:47 VanessaE inb4 "REALLY fix FacePositionCache data race"  ;)
22:47 VanessaE how do you even tell which thread is the crash?
22:48 VanessaE (I mean from the backtrace)
22:48 TBC_x sometimes the threads are touching the same data
22:53 VanessaE well what I meant was, what led you to assume thread #4 in this case/
22:54 VanessaE what in these backtraces gives away the actual crash versus all the other crap that just happened to be running at the same time?
22:56 VanessaE wait, I think I see.
22:56 VanessaE [Switching to Thread 0x7ffff1acd700 (LWP 17484)]  -->  Thread 4 (Thread 0x7ffff1acd700 (LWP 17484)):
23:01 bluegreen joined #minetest-dev
23:07 bluegreen joined #minetest-dev
23:18 sloantothebone How do I make a node swimable?
23:19 TBC_x but the thread got its lock
23:19 sloantothebone How do I make it act like a water_source
23:20 TBC_x theoretically, It may be influenced by data races in other parts of the code
23:21 VanessaE sloantothebone: keep your modding questions to #minetest!
23:24 VanessaE TBC_x: that was my first thought actually.
23:24 VanessaE but just *where* is a different matter :)
23:33 TBC_x which thread is supposed to call RemoteClient::SetBlocksNotSent?
23:53 paramat now pushing game 598 604 605 606

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