Minetest logo

IRC log for #minetest-dev, 2015-09-14

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

All times shown according to UTC.

Time Nick Message
00:24 Mallot1 joined #minetest-dev
00:41 twoelk joined #minetest-dev
01:31 twoelk|2 joined #minetest-dev
01:40 HulkSmash joined #minetest-dev
02:08 est31 joined #minetest-dev
02:50 leat joined #minetest-dev
03:22 Siva_Machina joined #minetest-dev
04:17 est31 can I push https://github.com/est31/minetest/commit/8e83ca4b624a9bcc49087c126993c23504d1a97a
04:31 asl97 joined #minetest-dev
04:38 est31 did a PR
04:46 hmmmm looks good to me
04:47 H-H-H joined #minetest-dev
05:21 est31 btw wheather has been FULLY removed, right?
05:22 est31 I mean: do we still need the hack of commit 06cdce1e12?
05:22 est31 I mean 06cdce1e12
05:28 est31 btw why has it been removed?
05:36 OldCoder Has mapformat.txt been deleted?
05:36 OldCoder A vanilla build just failed
05:36 OldCoder file INSTALL cannot find
05:36 OldCoder "/opt/minebest/src/core/server/minetest/doc/mapformat.txt"
05:36 est31 ahhh
05:37 est31 sorry my fault
05:37 est31 OldCoder, yes it has been renamed to worldformat.txt
05:37 est31 because it not just documents the map's format
05:37 est31 but of the whole world
05:38 OldCoder But the build fails?
05:38 OldCoder Generic make && make install gives that error
05:38 OldCoder file INSTALL cannot find
05:38 OldCoder "/opt/minebest/src/core/server/minetest/doc/mapformat.txt"
05:38 OldCoder ?
05:38 est31 well surely it does have to be fixed
05:39 OldCoder So, a recent change?
05:39 OldCoder Nobody but me has compiled recently? Odd
05:39 est31 yes
05:39 est31 well I have compiled but I never do make install
05:40 OldCoder Heh
05:40 OldCoder Well, my bug report is submitted
05:40 OldCoder Presently, MT can't be built
05:41 * OldCoder patches it temporarily by cp -p doc/worldformat.txt doc/mapformat.txt
05:42 OldCoder
05:42 Hunterz joined #minetest-dev
05:42 OldCoder BTW VE says mapgen7 does not work with the types of lighting in moretrees
05:43 OldCoder Does anybody know if that is corrected?
05:43 est31 perhaps you have luck and paramat appears
05:43 est31 he's on github at least
05:44 OldCoder One step at a time
05:44 OldCoder Have recently tried mg7 and seems to work
05:47 est31 OldCoder, can you look whether I've done a typo : https://github.com/est31/minetest/commit/b3682e7dae87580c91de955ced90c574ddf56925
05:50 OldCoder Hi
05:50 OldCoder est31, Reviewing
05:50 paramat joined #minetest-dev
05:51 paramat can you link me to 06cdce1e12 ?
05:51 OldCoder paramat, HI
05:51 OldCoder est31, let me repull the code
05:51 OldCoder Moment
05:51 paramat moretrees still has shadow bugs, more common in mgv7, not fixed
05:52 OldCoder Hi; thanks!
05:52 paramat weather was proller's weather and had many issues
05:52 paramat (dynamic weather)
05:53 OldCoder ./CMakeLists.txt:install(FILES "doc/mapformat.txt" DESTINATION "${DOCDIR}")
05:53 OldCoder est31, that is in the code; odd, your change is not present
05:53 OldCoder I see no typo
05:53 est31 ok pushed
05:53 OldCoder The change simply isn't there
05:53 OldCoder Ah
05:53 OldCoder This is new
05:53 OldCoder Repulling
05:55 OldCoder est31, rebuild works
05:55 OldCoder t y
06:08 kaeza joined #minetest-dev
06:20 est31 really, to be honest, the serialisation version is the totally wrong place to send wheather data
06:20 est31 to indicate that
06:20 est31 I mean
06:21 est31 and the serialized mapblock is the wrong place to send the wheather data
06:21 est31 yes, it fits with the client's memory structure
06:21 est31 but that sholdn't be used to design a protocol
06:23 est31 or not?
06:23 est31 I'm not sure
06:24 paramat i'll remove those commented-out lines then merge #3175
06:24 ShadowBot https://github.com/minetest/minetest/issues/3175 -- Firelike drawtype: Improve code by paramat
06:24 est31 okay
06:29 paramat hm actually i can simplify it more by replacing hOffset and hOffsetBot by literals, those were probably variables only for easy adjustment of the drawtype
06:30 est31 yea its better to have variables
06:30 est31 then you can tune easier
06:30 est31 and make less mistakes
06:32 est31 well I guess ill leave the wheather serialisation as it is right now
06:33 hmmmm weather data has not been part of minetest for quite some time now
06:33 hmmmm when we got rid of proller's shit
06:34 est31 it still gets sent
06:34 est31 hardcoded to 0
06:34 hmmmm ugh
06:34 est31 with a comment // deprecated
06:35 est31 look at MapBlock::{DeS,s}erializeNetworkSpecific methods
06:35 est31 they were added by proller's commit
06:37 est31 still I wonder what the actual reason was why we have no wheather
06:37 est31 anymore
06:37 paramat i was just imagining the season changing and all the snow on a mountain melting to water =s
06:39 hmmmm if minetest had stayed in that direction, the season would change and half the snow on the mountain would melt to water, before crashing
06:39 hmmmm and then the water would probably get stuck or something
06:39 red1 joined #minetest-dev
06:41 est31 yea its a nightmare to get optimized
06:42 est31 imagine all loaded blocks being modified by the server all the time
06:42 est31 just because of seasons
06:42 hmmmm lol, not even that
06:42 hmmmm last time i checked, proller's finite liquid is STILL buggy
06:43 nrzkt joined #minetest-dev
06:44 est31 the problem here is that we have to do a very unrealistic simulation of nature
06:44 est31 one that doesnt take resources
06:44 est31 in nature however, there is tons of "resources" spent on tons of things
06:51 blaze joined #minetest-dev
07:02 paramat now merging 3175
07:03 Darcidride joined #minetest-dev
07:07 paramat complete
07:26 paramat ~tell sfan5 i'm going through some old PRs for mtgame, are 2 '-1's enough to close an old PR? such as game#165 game#178
07:26 ShadowBot paramat: O.K.
07:32 paramat ~tell sfan5 perhaps it should be 3 '-1's as that's a majority. perhaps you could add your vote to the threads i am updating?
07:32 ShadowBot paramat: O.K.
07:34 nrzkt joined #minetest-dev
07:41 est31 game#165
07:41 ShadowBot https://github.com/minetest/minetest_game/issues/165 -- Add moonflower mod by RealBadAngel
07:41 est31 game#178
07:41 ShadowBot https://github.com/minetest/minetest_game/issues/178 -- Add tin by Zeg9
07:43 est31 the real big problem of VAEs is lighting
07:44 est31 we can't update our per-node lighting maps every time a VAE moves
07:46 rubenwardy joined #minetest-dev
07:47 est31 I mean VAEs should influence each other
07:47 est31 when you place a torch at a spaceship or so
07:47 est31 lol a flying problem
08:00 Yepoleb_ joined #minetest-dev
08:07 rubenwardy what about collision, est31? That would require a lot of map look ups.
08:10 est31 thats a problem as well
08:12 celeron55 collision isn't impossible at all, but there's not much room for sloppy design or coding in it
08:12 celeron55 any special lighting has to wait for a huge lighting rework; it's useless to even think about that in the current system
08:14 est31 any ideas how the new system could look like
08:14 celeron55 buildat has VAEs and it uses bullet (through urho3d's wrapping of it) for their physics collisions
08:15 leat joined #minetest-dev
08:16 celeron55 i don't know; lighting is problematic because people are used to how lighting in minetest works and for example switching to more traditional 3d lighting would really put people off
08:17 celeron55 you need something special to make caves work as well as they do now
08:18 celeron55 and houses and whatever
08:18 waressearcher2 what is VAE ?
08:18 rubenwardy Voxel Area Entities
08:18 rubenwardy see
08:18 rubenwardy ~dev todo
08:18 ShadowBot TODO - Minetest Developer Wiki -- http://dev.minetest.net/todo
08:21 celeron55 basically to get it right we'd need something that is called "global illumination"
08:21 waressearcher2 there is 0.5.0 coming soon ?
08:22 est31 no idea
08:22 rubenwardy versions don't matter that much, waressearcher2
08:23 celeron55 (minetest kind of weirdly has global illumination in its current voxel-based ligting, believe it or not)
08:23 nrzkt waressearcher2: 0.4.13 has been released 1 month ago, wait :)
08:24 waressearcher2 est31: its just I'm reading http://dev.minetest.net/TODO#Big_protocol_changes "The point of this section is to list such features so that they can be merged at once: a one-time incompatibility, perhaps coinciding with the release of 0.5.0, is better than breaking compatibility all the time"
08:25 waressearcher2 I thought its like "embrase yorselves 0.5.0 is coming"
08:26 est31 yea for a time this was how things looked like
08:26 est31 then sb tried to implement breaking changes some people vetoed
08:26 celeron55 it's more like "embrace yourselves, this will take forever to happen"
08:26 est31 now we regard backwards compat as something very important
08:27 est31 lol
08:28 est31 but if there is really a feature that needs a breaking change... why not then
08:28 est31 but right now I haven't seen any feature that was worth it
08:28 est31 and which really needs that change
08:30 waressearcher2 how about a clean slate ?
08:30 rubenwardy as long as there are world convertors, and mods still work, it's fne
08:30 rubenwardy waressearcher2, a rewrite?
08:30 rubenwardy do you realise how much work that would be?
08:31 waressearcher2 right, with all that experience, new fresh start will be better
08:31 est31 there are tons of half finished minecraft clones out there
08:31 rubenwardy ~g hexahedra
08:31 ShadowBot rubenwardy: https://en.wikipedia.org/wiki/Hexahedron | A hexahedron (plural: hexahedra) is any polyhedron with six faces. A cube, for example, is a regular hexahedron with all its faces square, and three squares ...
08:31 rubenwardy hexahedra.net
08:31 rubenwardy it's a project that apparently is based on the idea of Minetest
08:31 rubenwardy [citation needed]
08:32 rubenwardy anyway, this isn't really #minetest-dev quality
08:32 kahrl_ waressearcher2: http://www.joelonsoftware.com/articles/fog0000000069.html
08:36 paramat yeah i prefer organic evolution of code, mgv7 was almost scrapped but now look at it
08:37 est31 actually I am really impressed by their mapgen
08:37 est31 https://www.youtube.com/watch?v=_uSFLOPTVdE
08:37 est31 those mountains
08:37 rubenwardy have you seen their mapgen specification language?
08:37 est31 no
08:38 rubenwardy 1 sec
08:38 book` joined #minetest-dev
08:38 celeron55 that project seemed to die about a year ago
08:39 rubenwardy http://forum.hexahedra.net/t/development-for-0-2/41/13
08:40 rubenwardy no commits since the 4th of December
08:40 rubenwardy https://github.com/Nocte-/hexahedra/
08:41 paramat for a moment i thought it had voxel-resolution realtime shadows, i like the idea of that
08:41 celeron55 my problem with hexahedra is that it's even more less clear what it's trying to do than in minetest's case
08:41 est31 yea
08:41 celeron55 it's less of a game, and it's less extendable than minetest
08:41 celeron55 so what is it then
08:42 rubenwardy a wip
08:42 celeron55 yes, but what's the goal
08:42 rubenwardy no idea
08:43 kahrl_ it's more infinite than minetest at least :)
08:43 rubenwardy a Minetest-clone
08:43 rubenwardy less of a complicated mess, I guess ;)
08:43 est31 interesting
08:43 celeron55 that's just stupid in my opinion; minetest is already vague enough, why would anyone in their right mind clone minetest, lol
08:43 est31 Tesseract will like this
08:44 est31 their server is hexaedra-server
08:44 est31 not hexaedraserver
08:44 rubenwardy as a rewrite, probably
08:44 est31 minetest did some things right
08:44 est31 cubic mapblocks
08:44 est31 leading to limitless height
08:45 celeron55 hexahedra has a completely different Lua API, right?
08:45 est31 totally dynamic nodedefs (nothing hardcoded)
08:45 rubenwardy limitless, as in 31,000
08:46 est31 we have some uglier things though as well
08:46 rubenwardy The author has been developing the custom made noise library that Hexahedra uses, last at 22 Jul
08:47 est31 like what you say
08:47 rubenwardy Minetest's API isn't faultless
08:47 est31 that limit
08:47 celeron55 minetest is faulty has hell but at least some people find it useful
08:48 rubenwardy it's still being developed, and isn't really a working product yet
08:49 celeron55 the largest issue is that we, and i, don't actually understand why MT is as popular as it is and might just screw up everything like idiots 8)
08:50 Darcidride joined #minetest-dev
08:50 est31 not doing anything is bad as well :)
08:51 est31 I like minetest (and all minecraftlikes in general but I only have minetest experience) because it has so many variations. in an RTS you can only place buildings, in a shooter you can only shoot at people
08:51 est31 in minetest you can shoot at people while placing buildings
08:52 est31 well, the shooting isnt as good as in ego-shooters
08:53 celeron55 that's the "ridiculous sandbox" aspect, which i agree is a pretty strong point
08:53 est31 perhaps its only because we live in the 2010s
08:54 est31 its a thing that is popular for a short time
08:54 est31 then vanishes
08:54 celeron55 (i think i actually wrote a blog post about that a year or two ago)
08:54 est31 (regarding voxellikes)
08:54 est31 perhaps I've read that and my brain didnt store it with proper attribution
08:55 est31 stupid brain
08:57 celeron55 i'm referring to this http://c55.me/blog/?p=1491
08:57 celeron55 i don't entirely like what i wrote in that but it still has some merit to it in my opinion
08:58 est31 hexahedron is affero GPL 3
08:58 est31 O-o
08:58 celeron55 oh god why
08:59 paramat i like that post
09:03 waressearcher2 celeron55: why don't you like gpl3 ?
09:03 waressearcher2 torvalds also doesn't like it
09:03 celeron55 Affero GPL is not free enough
09:04 celeron55 LGPL is the least free i personally accept
09:04 nrzkt go to BSD licenced :D
09:05 est31 my thought to this is: if I want to make money with something, why should I not give a part of it to the original creator (you)?
09:06 est31 copyleft just separates the concepts of free software and proprietary software
09:07 est31 but LGPL is OK IMO, we are an engine
09:09 est31 and parts of minetest are even MIT licensed
09:19 rubenwardy by free do you mean permissive?
09:20 est31 me? or c55?
09:23 rubenwardy <celeron55> Affero GPL is not free enough
09:23 waressearcher2 est31: "should I not give a part of it to the original creator", because some people just like Steve Jobs ?
09:23 est31 yeah, thas his definition. I can understand him
09:24 est31 what does this have to do with steve jobs?
09:24 waressearcher2 est31: he lied to wozniak about bonus money
09:26 celeron55 rubenwardy: in my head free and permissive are quite the same things
09:27 rubenwardy okay
09:27 rubenwardy It's just that GNU defines free as copyleft, which is confusing
09:27 celeron55 but there might be some language barrier in there
09:27 rubenwardy *FSF
09:28 celeron55 oh, that too; i don't really listen to anything FSF says
09:38 est31 btw about the VAE phyics implementation in buildat
09:39 est31 how do you give the VAE's shape to bullet?
09:42 celeron55 buildat constructs a bullet object thingy (i don't remember what they are called) and just adds a lot of boxes to it
09:42 celeron55 not one box per node but it optimizes them somewhat
09:43 celeron55 the whole ground in it is actually just VAEs put next to each other
09:46 leat joined #minetest-dev
09:49 celeron55 est31: http://i.imgur.com/MHCP2UP.jpg
09:49 T4im joined #minetest-dev
09:52 est31 cool
09:54 celeron55 those physics get kind of weird if you for example fall really fast into the ground
09:55 celeron55 you might fall into the ground and get weirdly squeezed in between of some boxes
09:55 celeron55 then you have to walk and jump and wiggle yourself out from there 8)
09:58 leat joined #minetest-dev
09:59 waressearcher2 celeron55: is that screenshot from minetest ?
09:59 celeron55 no, it's from this crazy experiment https://github.com/celeron55/buildat
10:01 est31 I think a good goal for minetest would be to be modular
10:02 est31 e.g. what zeno does
10:02 est31 (that we can use multiple renderers)
10:02 est31 not as modular as buildat
10:03 est31 also not as generic as it
10:09 paramat left #minetest-dev
10:09 est31 also, we should federate minetest's data formats
10:09 est31 minecraft is the big player here with extra wikis created just by people who create minecraft compatible minecraft clones
10:10 est31 but I think devs should adopt minetest's formats because they are better in many regards
10:11 est31 minecraft is very simple compared to us
10:11 est31 but yeah, perhaps they are the HTTP/1.0 and we are the HTTP/1.2. who implements us as a hobby project xD
10:17 celeron55 minetest's formats aren't that great, but if they are revised some day, then i imagine people might want to use them
10:17 est31 what do you think are problems
10:18 celeron55 they are full of legacy crap
10:18 est31 yea there is some of it agreed
10:36 est31 most from the days there was no mt_game
10:39 OldCoder celeron55 I have observed a puzzle for weeks. If ABMs have the same chance, they seem to trigger often on exactly the same nodes. But the C++ code seems to iterate through the ABM list and call myrand for each one...
10:40 OldCoder How is it that one node gets triggered a bunch of times simultaneously?
10:46 leat joined #minetest-dev
10:47 celeron55 OldCoder: based on the information you just gave, what you said isn't possible
10:47 OldCoder All right; will continue to debug... but...
10:48 celeron55 i don't remember anything that could cause that
10:48 OldCoder What I've observed is that mobs are created by ABMs... even ABMs that are unrelated end up with them clumped together
10:48 OldCoder Will look again at the Lua code
10:48 OldCoder But this is highly consistent
10:48 OldCoder It is not visible in most worlds because
10:49 OldCoder People set active object count limit low
10:49 OldCoder If the limit is increased, the issue becomes visible
10:49 OldCoder Based on what you just said, the issue must be somewhere at the Lua level, but I've traced it down
10:49 celeron55 oh, well that could be for example because the iteration order is constant, and the ABM implementation maybe doesn't add a mob if there already are many, in which case the mobs end up in the beginning part of the iteration?
10:49 OldCoder One moment; it is this line of C++ code that is of interest:
10:50 OldCoder for(std::vector<ActiveABM>::iterator
10:50 OldCoder i = j->second.begin(); i != j->second.end(); ++i) {
10:50 OldCoder if(myrand() % i->chance != 0) continue;
10:50 OldCoder 3 lines, excuse me
10:50 OldCoder Doesn't this mean that each ABM gets a separate myrand call?
10:51 OldCoder Note that I'm referring to multiple distinct ABMs
10:51 celeron55 yes; and i do not believe there are any such problems on the C++ side
10:51 OldCoder triggering at the same time on the same nodes
10:51 celeron55 oh, well that can happen of course
10:51 OldCoder Well, I'll continue to trace through the Lua code, then. But different ABMs, different mods, are triggering up to 64 times at once on the same node
10:51 celeron55 the nodes are iterated only once, right?
10:51 OldCoder Hm?
10:51 cvtsx joined #minetest-dev
10:52 OldCoder Yes but each gets its own C++ myrand call, right?
10:52 celeron55 yes but that doesn't mean they couldn't activate at the same time, especially if they have a relatively high chance and a big interval
10:52 OldCoder They could, yes; but they activate in clumps spread far apart
10:53 celeron55 i believe people usually use too big intervals and and too big chances in their ABMs
10:53 celeron55 -and
10:53 OldCoder Yes; but even so, there should not be clumps. Go on.
10:53 celeron55 well, dunno then
10:53 OldCoder Comment on too-big chances, do you mean chance too large?
10:53 OldCoder I.e. low odds?
10:53 celeron55 too high odds
10:53 OldCoder All right. No, I have things at chance 50000
10:54 OldCoder Interval 30
10:54 celeron55 okay that's more or less intended usage
10:54 OldCoder And they still appear in clumps. Almost certainly a bug at some level. I will report back.
10:54 OldCoder The thing is, the ABMs go right down to the minetest.register_abm level
10:54 OldCoder So there isn't much Lua to debug
10:55 celeron55 wait, chance of 50000...
10:55 OldCoder Is there a limit of 32767 or something?
10:55 OldCoder The myrand code seems to support unsigned long long arithmetic
10:55 celeron55 isn't the return value of myrand() something like int16 or uint16
10:55 OldCoder Checking; moment
10:56 OldCoder No it is u32
10:56 OldCoder There was a known bug in the past
10:56 OldCoder Where range size could not be too large for PseudoRandom
10:56 OldCoder But myrand seems to support u32
10:56 OldCoder Not enough for you to comment further, probably
10:57 OldCoder I can add some printf's to the C++ code
10:57 OldCoder and see if it is in the myrand code or a different level
10:57 celeron55 well the C standard rand() returns int but the reutrnvalue is actually 0...RAND_MAX and RAND_MAX is often 32767
10:57 celeron55 i don't even know what this PcgRandom::next() exactly does; it's not documented properly
10:57 OldCoder Reviewing code
10:58 OldCoder It is an attempt at a higher quality pseudo-random generator
10:58 celeron55 that i know
10:58 OldCoder Hm; I could replace it in my copy and see
10:58 celeron55 just add a dstream<<myrand()<<std::endl somewhere, lol :P
10:59 OldCoder Right in the ABM code
10:59 OldCoder is where that would go
10:59 OldCoder Got to sleep now; I'll comment if this gets further
10:59 OldCoder Zzz
11:16 Elinvention joined #minetest-dev
11:22 blaze joined #minetest-dev
11:35 Darcidride_ joined #minetest-dev
11:38 kahrl joined #minetest-dev
12:00 kilbith joined #minetest-dev
12:01 kilbith left #minetest-dev
12:14 Darcidride__ joined #minetest-dev
12:23 DFeniks joined #minetest-dev
12:23 rubenwardy joined #minetest-dev
12:39 Darcidride__ joined #minetest-dev
12:59 AnotherBrick joined #minetest-dev
13:11 asl97 joined #minetest-dev
13:23 est31 joined #minetest-dev
13:35 RealBadAngel joined #minetest-dev
13:39 rubenwardyy joined #minetest-dev
13:43 rickmcfarley joined #minetest-dev
13:53 rickmcfarley joined #minetest-dev
14:05 Megaf Hi celeron55, what can you say about Minetest on Wayland?
14:06 Megaf Minetest is the only reason I'm using X.org by the way.
14:14 zupoman joined #minetest-dev
14:20 BrandonReese joined #minetest-dev
14:21 T4im Megaf: I'm running it on xwayland just fine
14:22 Megaf T4im: how do you do it? Here it doesnt pointer lock or somehting like that, pointer acts like an analog joystick
14:23 T4im I don't do anything special.. it just worked
14:24 T4im (keep in mind: xwayland, not _just_ wayland)
14:27 Megaf ok, so how you run it? Here I start GNOME on Wayland, then I try to run Minetest
14:27 Megaf so is wayland
14:28 T4im I run it just like that, too
14:28 T4im gnome under (x)wayland running minetest
14:30 T4im xwayland is the compatibility layer,.. basicly a x server on top of wayland.. that's why most stuff should just work out of the box
14:32 H-H-H joined #minetest-dev
14:37 CraigyDavi joined #minetest-dev
14:51 celeron55 Megaf: i don't know much about wayland, not being a user of it myself (on desktop that is; my phone does use it though 8))
14:51 celeron55 but i figure it's an irrlicht problem
14:52 celeron55 assuming your wayland compositor isn't buggy
14:59 DFeniks joined #minetest-dev
15:04 CraigyDavi joined #minetest-dev
15:10 rickmcfarley joined #minetest-dev
15:43 hmmmm joined #minetest-dev
16:07 Hunterz joined #minetest-dev
16:29 Robert_Zenz joined #minetest-dev
16:33 Elinvention joined #minetest-dev
16:37 nrzkt joined #minetest-dev
17:01 everamzah joined #minetest-dev
17:05 proller joined #minetest-dev
17:14 leat1 joined #minetest-dev
17:16 leat1 joined #minetest-dev
17:19 MinetestForFun joined #minetest-dev
17:26 leat1 joined #minetest-dev
17:37 leat1 joined #minetest-dev
17:41 ElectronLibre joined #minetest-dev
17:42 Calinou joined #minetest-dev
17:53 kilbith joined #minetest-dev
17:56 Krock joined #minetest-dev
17:57 leat1 joined #minetest-dev
18:12 red1 joined #minetest-dev
18:12 red1 joined #minetest-dev
18:15 HulkSmash joined #minetest-dev
18:16 Taoki[mobile] joined #minetest-dev
18:34 Elinvention joined #minetest-dev
19:13 * VanessaE sighs
19:13 * sfan5 stares
19:14 VanessaE are you guys ever gonna fix the bugs that happen when entities end up out of world bounds?
19:14 VanessaE (such as corrupt blocks, huge FPS loss, server crashes)
19:14 VanessaE now even my FreeForAll just got bit
19:14 nrzkt why did you go to 32XXX,32XXX,32XXX ? :p
19:15 VanessaE to chase down a bad block error
19:15 VanessaE found a fuckton of entities falling through the void
19:15 VanessaE (MOBs in this case)
19:15 nrzkt void paths are immersive :D
19:15 nanepiwo note: trying something like /area_pos1 99999999,99999999,99999999 would try to create the entity in this void and crash the server as well...
19:16 VanessaE nanepiwo: correct.
19:16 nrzkt oh, this is great bug !
19:16 nrzkt why is this value allowed ? core ? area mod ?
19:16 VanessaE forget why it's allowed.
19:16 VanessaE why is the engine crashing instead of just deleting what ends up there?
19:17 nrzkt engine should forbid those operations, not deleting . Stopping everything but don't crash
19:17 VanessaE it should delete, too.
19:17 VanessaE falling entities should not just "pile up" at the edge.
19:18 VanessaE also it causes a huge memory leak as well.
19:18 VanessaE see my graph: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/stats-daily.html
19:19 VanessaE FFA was using around 2.7 GB by itself.  operating totally normally, save for killing my client's FPS once the entities all loaded up
19:19 VanessaE crashed when I tried to select the affected area with worldedit (forgot about that crash source)
19:20 VanessaE the bad block coords were 1932,-79,-1933
19:20 VanessaE so right at the edge of the world.
19:22 nrzkt it's not world border
19:22 VanessaE yes, it is.
19:22 nrzkt on mu server /teleport 10000 50 10000 works perfectly fine, border is 32XXX
19:22 VanessaE those are block coords
19:22 VanessaE not node coords.
19:23 VanessaE 1932*16 -> 30912
19:23 nrzkt k
19:23 asl97 VanessaE: 16 is the chunk size isn't it?
19:23 VanessaE yes
19:23 VanessaE (well it's the mapblock size, I think a "chunk" is 5x5x5 mapblocks)
19:25 asl97 ah yea, iirc, there no way to change the mapblock size through the conf
19:25 VanessaE in ANY case, it's clearly unsafe to let anyone or anything get near the map edges.
19:27 VanessaE here's the backtrace from the crash that worldedit caused when I tried to select the area:  http://pastebin.ubuntu.com/12411552/
19:27 VanessaE (in worldedit, you can select an area and do stuff like //clearobjects to just operate in that area)
19:28 Hijiri joined #minetest-dev
19:30 VanessaE meanwhile, I did a /deleteblocks in that area, radius 5 and then did a global clearobjects
19:30 asl97 how does it fall all the way down though
19:31 VanessaE they looked like sand monsters (from carbone_mobs); there is a long, vertical shaft going from there on up past where I could see, so they probably spawned in a surface desert and then fell down at some point.
19:33 VanessaE and yes, somehow those mobs managed to *punch* me while both they and I were floating in the out-of-bounds void.
19:36 VanessaE and furthermore, my client crashed when it tried to display the server shutdown/crash screen.
19:44 VanessaE plus to top it all off, I'm starting to see those "suspiciously large amount of entities in block" and "block already has more than X objects, forcing detele"
19:45 Player_2 joined #minetest-dev
19:45 VanessaE with object counts into the 70'000 range
19:46 VanessaE ...forcing delete*
19:56 leat1 joined #minetest-dev
20:09 zat joined #minetest-dev
20:27 OldCoder VanessaE, //deleteblocks won't delete entities, right? Only solution is /clearobjects ?
20:28 OldCoder Is there a way to do clearobjects just for the region below -32000 ?
20:44 ElectronLibre left #minetest-dev
20:46 red1 joined #minetest-dev
20:46 Taoki[mobile]_1 joined #minetest-dev
21:04 Wayward1 joined #minetest-dev
21:06 Fritigern joined #minetest-dev
21:13 7GHAA60XN joined #minetest-dev
21:28 hmmmm /deleteblocks does delete any entities statically stored in those blocks
21:28 hmmmm no objects that are currently active on the map however
21:34 everamzah yeah that's a bummer
21:53 paramat joined #minetest-dev
22:05 paramat now merging game#442 and game#601
22:05 ShadowBot https://github.com/minetest/minetest_game/issues/442 -- Change obsidian glass drawtype to glasslike_framed_optional by MT-Modder
22:05 ShadowBot https://github.com/minetest/minetest_game/issues/601 -- Remove beds/Changelog.txt & farming/API.txt by CraigyDavi
22:12 paramat complete
22:32 Fritigern joined #minetest-dev
22:34 Yepoleb left #minetest-dev
22:42 paramat left #minetest-dev
23:16 asl97 joined #minetest-dev

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