Minetest logo

IRC log for #minetest-dev, 2014-10-05

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

All times shown according to UTC.

Time Nick Message
00:54 MikeFair joined #minetest-dev
01:02 paramat joined #minetest-dev
01:11 Taoki joined #minetest-dev
01:20 ShadowNinja c55's never voiced, not sure why.  I guess he chose that.
01:22 ShadowNinja 8 lights per scene node sounds like a ridiculously low limit.  Are you sure we're doing this right?  Maybe every node should be a scene node, but I'd think that would be very slow.
01:23 ShadowNinja We could create scene nodes for every light source though (or combine them with a little extra logic).  That shouldn't be too slow.
01:23 ShadowNinja Unless 8 per scene node means 8 affecting the scene node.
01:56 ShadowNinja Also, how should I go about changing the DB schema?  I'll have to record a version somewhere.  maybe sqlite3-schema-version = 1 in map_meta.txt?
01:57 ShadowNinja Seperate x, y, and z coordinates would let SQLite3's optimizer optimize queries better.
02:25 hmmmm no, dammit!
02:25 hmmmm i thought we decided we're going to generate lightmaps on the fly
02:26 Zeno` joined #minetest-dev
02:26 hmmmm dynamic lights should be used very sparingly - like for a torch the player is holding
02:28 Miner_48er joined #minetest-dev
02:47 paramat i think the plan is hardware lighting for sun/moon only, because 8 per scenenode is obviously too few, even with 1 scenenode per mapblock
02:49 VanessaE how about poking at RealBadAngel instead of guessing? :P
02:52 paramat hmmmm, want to be the new audiovisual manager? RBA needs someone who is around when he is, see https://github.com/minetest/minetest/issues/1710
02:53 paramat my first pull request for a very long time https://github.com/minetest/minetest_game/pull/328 (for minetest game devs)
03:25 alexxs joined #minetest-dev
03:28 rickmcfarley joined #minetest-dev
04:01 kaeza joined #minetest-dev
04:10 hmmmm meh, i'm not going to make promises of being around in any capacity... i just hope to get some commits in because there's a lot of stuff on my own backlog, that's all
04:12 hmmmm I'm still weary about adding NodeResolver to INodeDefManager.
04:44 RealBadAngel what i want is another coder that will actually work on the stuff with me
04:45 RealBadAngel c55 is a great coder but pure virtual, he does nothing but complain
04:46 RealBadAngel actually its a way easier to complain than to code something
04:46 RealBadAngel maybe a middle age crysis?
04:47 RealBadAngel "lets ""talk"" about sex"
04:48 RealBadAngel voila
04:48 RealBadAngel here he comes
04:49 celeron55 you should note that nick changes are global to freenode
04:49 RealBadAngel i know
04:49 celeron55 i didn't come to this channel
04:49 RealBadAngel i explained that to folks
04:50 RealBadAngel but i guess thats not the point
04:50 celeron55 also, i don't complain a lot
04:51 celeron55 i mainly complain when someone pings me and asks me to complain
04:51 celeron55 which happens rarely now
04:51 RealBadAngel you do when it comes to my code
04:52 celeron55 no i don't, that's just bullshit
04:52 celeron55 i did at some point but now you're just recycling your old memories
04:52 RealBadAngel not so old
04:53 RealBadAngel but lets leave that behind
04:54 RealBadAngel as i said previously, i need some1 to work with me
04:54 celeron55 altough, "he does nothing but complain" might be valid as even while i do it little, i don't do anything else 8)
04:55 RealBadAngel so wake up buddy
04:55 RealBadAngel theres lotsa to do
04:56 RealBadAngel ive spent lots of time learning irrlicht
04:56 RealBadAngel and i know how to make things work
04:58 RealBadAngel youre the guru here anyway, who knows, maybe bill still have some spare billions just to buy you out ;)
05:00 RealBadAngel but definitely he wont pay if u will be doing nothing
05:02 RealBadAngel but seriously, we are about to have biggest changes since 0.4
05:03 RealBadAngel meshes and lighting are huge changes that require massive code changes
05:03 RealBadAngel im not able to change it all even in the series of small pulls
05:04 sol_invictus joined #minetest-dev
05:04 RealBadAngel single adding the mesh drawtype require MANY changes
05:05 RealBadAngel only you and kahrl here are able to comment it
05:05 RealBadAngel SN will be able to comment the style and whitespaces
05:05 RealBadAngel sapier seems to be on vacations
05:06 RealBadAngel rest dont know what im doing
05:06 hmmmm hrm
05:06 RealBadAngel yes, including you
05:07 hmmmm celeron55, so would you get upset if I added something to INodeDefManager?
05:07 RealBadAngel sorry to say that but we are highly specialized
05:07 RealBadAngel sometimes its good, sometimes bad
05:08 hmmmm I don't know much about Irrlicht, but I'm willing to learn to get things accomplished
05:09 RealBadAngel hmmmm, as they say, man learn whole the life. and dies stupid.
05:09 hmmmm I sorta gave up on the mapblock-as-a-scenenode thing because I found out that Minecraft used hardware occlusion queries and it sucked
05:10 RealBadAngel i also had "funny" ideas and gave up
05:10 hmmmm does minetest already have support for a wireframe mode?
05:10 VanessaE not directly, no.
05:10 hmmmm meh trivial to change
05:11 RealBadAngel why would we need that?
05:11 hmmmm setMaterialFlag -> irr::video::EMF_WIREFRAME
05:11 hmmmm to see how much we really are overdrawing
05:11 RealBadAngel no need to
05:11 VanessaE hmmmm: ShadowNinja I think showed that recently, or someone did
05:11 RealBadAngel there are built in methods to reduce the vertices
05:11 VanessaE no, wait, not SN
05:11 ShadowNinja hmmmm: Note that no-cliping disables occlusion culling.
05:11 VanessaE one of the new guys did
05:12 ShadowNinja VanessaE: Nah, that was me.
05:12 VanessaE oh ok
05:12 VanessaE wait a minute, wtf are you doing up this late? :)
05:12 celeron55 hmmmm: dunno, for what purpose?
05:12 * ShadowNinja hides
05:13 VanessaE ShadowNinja: that reminds me, how far did you get with that guy's culling algorithm?
05:13 hmmmm you remember those resolveNodeNames() that many of my register-nodes-in-lua-then-do-mapgen-thing features have
05:13 * celeron55 is early, not late 8)/
05:13 hmmmm well I got rid of those and made a centralized NodeResolver class
05:13 hmmmm naturally I'd put it into nodedef, but you have it set up as an abstract interface
05:14 RealBadAngel celeron55, and as usual. i talked too much, touched a wall. you are just avoiding the problems
05:14 hmmmm what other implementations of node def managers DID you intend to have anyway?
05:15 celeron55 RealBadAngel: i didn't even read what you said last; and yes, i am avoiding problems because that is the only way to keep the time down that i spend on minetest
05:16 ShadowNinja VanessaE: I haven't touched any culling code.
05:16 RealBadAngel celeron55, give the av to anybody else, sn, nore or blocken
05:16 VanessaE ShadowNinja: oh ok, then it WAS one of the new guys who was involved there.
05:16 RealBadAngel so i could work with somebody for christ sake
05:16 celeron55 hmmmm: it's my way of coding C++ that i have begun to use in the past few years where i make an interface to anything that is remotely complex so that implementation details have no chance of getting into header files (and headers stay simple with little dependencies)
05:17 celeron55 i've found this a very good way to do things
05:18 celeron55 of course, in some cases there can be many implementations (they are obviously easy to add anytime when needed)
05:19 hmmmm something something premature abstraction
05:20 hmmmm celeron, http://sebastiansylvan.com/2013/0​8/16/the-perils-of-future-coding/
05:20 celeron55 it makes coding much nicer because i can freely inline the whole implementation class
05:20 RealBadAngel licking an icecram through glass
05:20 celeron55 i've always hated C++'s class implementation syntax
05:20 RealBadAngel either something is fast and designed to do things or it is universal and worth shit
05:21 celeron55 so it's easier to code this way with less repetition
05:21 hmmmm well
05:21 celeron55 so i'm actually with that article
05:21 hmmmm I coded like 4 things that repeat themselves in varying capacity
05:21 RealBadAngel and so called flexibility and objects are just a mistake
05:21 RealBadAngel nowadays coders are "coders"
05:21 hmmmm it lets me see exactly which common elements there are
05:22 hmmmm like exactly what I'm doing here
05:22 RealBadAngel thats why m$ and intel are earning money
05:23 celeron55 i didn't understand what either of you just said
05:23 RealBadAngel before we had the very same platforms any only thing to change was the code
05:23 RealBadAngel and the style
05:23 RealBadAngel do the same thing but faster
05:23 RealBadAngel now you are talking bout headers
05:23 RealBadAngel rotfl
05:24 hmmmm meh, i was going to elaborate further, but i'd rather not interrupt another conversation to do so.
05:24 RealBadAngel count the cycles of your code
05:25 VanessaE celeron55: my husband likes to say that there are "coders", and there are "programmers".  The former just writes code just for the sake of writing code, without giving it much though whether that code works outside a very narrow set of parameters (e.g. it's paid for by some employer, who cares if an average Joe Sixpack can use it, etc).  And the latter is most of us here, people who want their code to form a cohesive proje
05:25 VanessaE ct that is useful and generally looks good to the average person.
05:25 celeron55 VanessaE: ok; so what is the point? i don't understand why RBA started talking about something like this
05:25 VanessaE I'm not sure I necessarily agree with his choice of the two words, but he sentiment is correct I think.
05:26 RealBadAngel celeron55, you dont know? im trying to push things around
05:26 RealBadAngel stop talkin bout abstracts
05:26 VanessaE what he's trying to say, I think, is that there's too much "let's just write some code just to write some code" and not enough "let's write some code that accomplishes a goal"
05:26 RealBadAngel do the code
05:26 celeron55 RealBadAngel: i don't care about anything you have said, and i was not talking to you but hmmmm
05:26 celeron55 hmmmm: could you please continue your discussion, this is fucking ridiculous
05:27 RealBadAngel sure
05:27 VanessaE (I'm done)
05:27 RealBadAngel as usual
05:27 hmmmm i think RBA is mad
05:27 RealBadAngel we cant code because you dont care :P
05:28 RealBadAngel yes im mad because im still trying to contribute
05:28 celeron55 RealBadAngel: can you point me to an issue that needs something from me instead of talking about philosophies
05:28 RealBadAngel celeron55, yes, we need to implement lights
05:29 celeron55 ok, what's the problem?
05:29 RealBadAngel simplify servers one to light level
05:29 RealBadAngel and add client side irrlicht real light sources
05:29 nore joined #minetest-dev
05:30 RealBadAngel problem is that a scene node can have 8 directional lights
05:30 hmmmm heh, it's ironic.  not meaning to be offensive, but i find that RBA often says more words than is necessary but he advocates coding the minimum that is necessary to get whatever the task is accomplished.
05:30 RealBadAngel so propably we need a hybrid system
05:31 celeron55 hmmmm: i find that even frustrating
05:31 celeron55 RealBadAngel: so what is your solution to that?
05:31 celeron55 8 directional lights per node is more than enough, but i'm guessing you can't change them per node when rendering
05:31 celeron55 8 per mapblock isn't enough
05:31 RealBadAngel hmmmm, celeron55 whats the point for pulling the big changes if even a few liners are blocked?
05:32 RealBadAngel celeron55, its a limit per scene node
05:32 hmmmm the solution isn't to make each mapnode a scene node.
05:32 celeron55 pulling big changes gives a much higher benefit for the amount of work required from the puller
05:32 hmmmm the solution is to dynamically generate a lightmap alongside the meshes and simply not use dynamic lights
05:32 celeron55 i.e. i can do it as a one-off thing, but i can't do many small pulls every day
05:33 RealBadAngel not when you need to constantly rebase the pulls
05:33 RealBadAngel lately im forced to rebase vs my own code
05:34 celeron55 assuming a clean implementation of this appears, i can handle the merging; what else do you expect?
05:35 RealBadAngel nothing more
05:35 celeron55 well, there are compatibility concerns but i am favourably biased now that i have had a computer for one year that can run anything you may come up with
05:35 rickmcfarley joined #minetest-dev
05:35 celeron55 so someone else has to push that concern if it still is a valid concern
05:35 OldCoder VanessaE, I'll discuss the off-topic definition that Abe made of a Coder with you later
05:35 VanessaE OldCoder: ok.
05:39 alexxs joined #minetest-dev
05:42 RealBadAngel celeron55, in a few days i will make a pull with meshes drawtype and converting nodeboexes into meshes
05:43 RealBadAngel when im done with meshes i will add the lights
05:44 RealBadAngel hope you will find time to review the changes
05:46 celeron55 i hope they are sane enough to pass my review
05:57 RealBadAngel i will try
06:32 nore joined #minetest-dev
06:37 hintss joined #minetest-dev
06:54 Hunterz joined #minetest-dev
07:06 paramat left #minetest-dev
07:10 Krock joined #minetest-dev
07:12 zat joined #minetest-dev
08:18 ImQ009 joined #minetest-dev
08:24 PenguinDad joined #minetest-dev
08:32 Calinou joined #minetest-dev
08:45 alexxs joined #minetest-dev
10:02 jin_xi joined #minetest-dev
10:28 edlothiol joined #minetest-dev
10:31 edlothiol left #minetest-dev
10:32 jin_xi hey nore can you help me understand camera_offset? i am trying to move particle handling more to irrlicht side but fail when more than 200 nodes away from origin
10:42 iqualfragile joined #minetest-dev
10:51 alexxs joined #minetest-dev
11:01 chchjesus joined #minetest-dev
11:06 BlockMen joined #minetest-dev
11:11 BlockMen would like to merge https://github.com/minetest/minetest/pull/1711
11:11 BlockMen comments?
11:18 troller joined #minetest-dev
11:27 Zeno` I am afraid to say
11:27 Zeno` there is still no early exit
11:30 Zeno` e.g. http://codepad.org/4GCX1fel
11:31 Zeno` Which is what hmmmm was suggesting I think
11:32 Krock BlockMen, you'll get a +1 from me
11:33 BlockMen Zeno`, i wrote my opinion on that already
11:34 Zeno` I couldn't see it
11:34 Zeno` nvm
11:34 nore jin_xi: everything is translated by camera_offset iirc
11:35 Zeno` All I can see is "Also, please modify the flow of parseColorString to "bail out early" if not hexstring, rather than encasing almost the entire body of the function in an if statement's block."
11:35 Zeno` Maybe I am blind
11:35 nore BlockMen, I like that
11:36 BlockMen Zeno`, its has to be done this way or the other way round for this: https://github.com/minetest/minetest/pull/1711/​files#diff-765e4fc5d8245871d9a5a4db834e4631R360
11:39 Zeno` That needs to be moved to line 318 then I guess
11:40 Zeno` meh, not that I care much :)
11:58 Zeno` Although, I'm not sure why you asked for feedback if you didn't want it
11:59 Bligh joined #minetest-dev
12:00 Amaz joined #minetest-dev
12:02 PilzAdam joined #minetest-dev
12:04 BlockMen Zeno`, since i said "want to merge" i was aiming mainly for core dev comments.
12:05 BlockMen but since im already working on that function: https://github.com/BlockMen/minetest/commi​t/0c49d505c7f6d638a36f04423b6290fa2006591c
12:06 BlockMen huos
12:06 BlockMen *ups
12:10 Zeno` not quite right but the idea is looking good
12:10 Zeno` e.g keeping your current flow the switch should probably be contained within an if
12:10 Zeno` i.e. only try a named colour if the hexpattern failed
12:17 Zeno` personally I'd make two new functions
12:21 CraigyDavi` joined #minetest-dev
12:23 Zeno` Since when can you use string values in a switch anyway?
12:26 BlockMen you cant, thats why i wrote ups :D
12:27 Zeno` oh :)
12:41 rickmcfarley joined #minetest-dev
12:51 PilzAdam in Java you can use strings in switch
13:02 Zeno` I've never been to Java
13:07 * sfan5 meows at PilzAdam
13:48 rubenwardy joined #minetest-dev
13:52 BlockMen updated https://github.com/minetest/minetest/pull/1711 again. travis build passed
13:52 BlockMen if noone claims anything im going to merge it in 30 minutes then
13:53 sfan5 "if noone claims anything im going to merge it in 30 minutes then"
13:53 sfan5 thats not how it works
13:54 Krock joined #minetest-dev
13:54 BlockMen nore agreed on it?
13:55 sfan5 he did? then it's ok
13:55 nore BlockMen, I agree on what it does, but I didn't test it, but the code looks fine
13:55 nore (well, I didn't read it very precisely either)
13:56 BlockMen hmm..maybe we should make write down that agree means "another dev has to write agree". then we prevent missunderstandings ;)
13:57 nore so to be clear: I agree on what it does, the code looks fine, but I didn't test it
13:57 BlockMen ok, then i change "will merge in 30 minutes" to "can someone check again and agree so i can merge?"
13:58 sfan5 BlockMen: see github comments
14:01 Megaf Hi all, is this a bug? http://paste.debian.net/plain/124590
14:02 sfan5 no it's feature
14:02 sfan5 why are you even asking
14:02 sfan5 it's obviously a bug
14:03 jin_xi aziz got quite the beating
14:04 kahrl what thread is that even, its ID does not show up in the debug stacks report
14:04 sfan5 probably the connectionthread
14:04 Megaf I'm asking if I should open open an issue or not
14:04 sfan5 it seems to be missing
14:06 Zeno` why doesn't parseColorString() call two other functions: parseHexPattern() or parseNamedColor()? It would be more readable and also enable getting rid of the unnecessary goto
14:06 kahrl does anyone else get the double free crashes etc that Megaf gets?
14:06 Zeno` kahrl, I have once
14:07 Zeno` haven't been able to reproduce it
14:07 kahrl nonetheless at the rate Megaf gets them I would suggest doing a memtest on that machine
14:07 Megaf several crashes that I have are hard to reproduce, sometime my server stays online for a week with no crashes
14:08 Megaf kahrl: I think no one leave minetestserver running for as long as I do
14:08 Zeno` the time it happened to me I was running valgrind, so perhaps it's time related... I don't know
14:08 kahrl anyone that runs a public server probably does
14:09 sfan5 83.165.141.1011/15   0/10.4.10, minetest, ?Cre Dmg PvP13.3d, 19.9d48, 100
14:09 sfan5 uptime 13.3d
14:10 sfan5 too bad we can't ask that person
14:11 BlockMen sfan5, everything changed
14:12 ImQ009_ joined #minetest-dev
14:12 sfan5 BlockMen: what about https://github.com/minetest/minete​st/pull/1711#discussion_r18435335 ?
14:13 BlockMen sfan5, blame github https://github.com/minetest/minetest/pull/1711/f​iles#diff-1305560bd8befb32862f0feeefabd02eR1239
14:13 sfan5 "ColorString:" doesn't really make sense because it looks like it's part of " Inventory location:"
14:13 sfan5 oh
14:13 sfan5 what about removing the :
14:13 BlockMen y?
14:14 sfan5 because it's like that everywhere else
14:14 Zeno` what about writing the code properly? If you're moving it why not fix it so that it's a bit more sane?
14:14 BlockMen there is no clear system in doc
14:14 ImQ009_ joined #minetest-dev
14:14 BlockMen or is?
14:14 Zeno` (I don't agree with functions that try to do so many things for no benefit)
14:15 Zeno` I know you didn't write the original code
14:16 sfan5 BlockMen: if you want to remove the : do that, if not you still have my approval
14:17 BlockMen sfan5, k i will remove it.
14:17 BlockMen also, two commits? one for [colorize and one for named colors?
14:17 BlockMen or "all in one"?
14:17 kahrl BlockMen: why are there only 5 named colors? what about yellow, cyan, magenta, ...?
14:18 sfan5 BlockMen: 2 commits is fine
14:18 BlockMen kahrl, and where do you want stop?
14:18 kahrl VGA palette or netscape palette
14:18 kahrl the VGA palette is probably enough
14:19 BlockMen wut?
14:20 kahrl err, I mean the 16 color palette
14:21 kahrl https://en.wikipedia.org/wiki/List_of​_8-bit_computer_hardware_palettes#CGA
14:21 kahrl either that or use the colors from the wool mod
14:22 kahrl I think the wool/dye mod colors would actually be best
14:24 BlockMen then i prefer the 8 color palette
14:24 Zeno` The whole function should really be as simple as: bool success; if (value[0] == '#') { success = parseHexColor(value, color); } else { success = parseNamedColor(value, color); } if (!success && !quiet) errorstream << "Invalid color: \"" << value << "\"" << std::endl; return success;
14:24 BlockMen *more
14:24 Zeno` No goto, no confusion, simple
14:25 BlockMen ok. how about that. i will make another pull for the color name stuff
14:25 BlockMen and just merge the [colorize part now
14:25 BlockMen then its clean sepperated
14:25 BlockMen and everyone can comment there what he thinks is best
14:29 Megaf Why dont you make something useful?
14:30 Megaf like track bugs now and optimze the code?
14:30 Megaf common
14:30 rubenwardy Easier said than done, really. Any open bug reports?
14:30 rubenwardy https://github.com/minetest/minetest/is​sues?q=is%3Aopen+is%3Aissue+label%3Abug
14:31 Zeno` I really don't understand why minetest makes things harder than they should be :(
14:32 Megaf This is a serious bug that happens on ALL platforms and you still have no clue on it https://github.com/minetest/minetest/issues/1512
14:32 Zeno` It's related to opening the "progress bar"
14:33 Zeno` Beyond that I don't know because the code is confusing
14:33 Zeno` It *may* even be an irrlicht bug...
14:35 Megaf And this is also a interesting bug, "a player is lost"
14:35 Megaf https://github.com/minetest/minetes​t/issues/1649#issuecomment-56878162
14:39 BlockMen ok, https://github.com/minetest/minetest/pull/1711/ is now [colorize only
14:41 BlockMen can i merge that now?
14:41 alexxs joined #minetest-dev
14:41 * Megaf gives thumbs down
14:42 kahrl BlockMen: include irrlichttyped_bloated instead of _extrabloated
14:42 kahrl -d +s
14:44 kahrl https://github.com/BlockMen/minetest/commit​/c8f2832fddb858ffcdeceda90525815640e60e8a#d​iff-d5a2988ef5cf71bce4a479a2e1b49426R1611 is indented too much
14:44 Zeno` I'd prefer it to be:  bool success; if (value[0] == '#') { success = parseHexColor(value, color); } else { success = false; } if (!success && !quiet) errorstream << "Invalid color: \"" << value << "\"" << std::endl; return success;
14:45 Zeno` Why the aversion to functions?
14:46 kahrl I agree with Zeno`
14:46 kahrl although I'd initialize success = false and remove the else
14:47 Zeno` either way
14:47 BlockMen i will do that in the next pull for the color names
14:47 BlockMen (zenos functions)
14:48 _Esteban joined #minetest-dev
14:49 GrimKriegor joined #minetest-dev
14:50 BlockMen kahrl, both comments of you fixed. can i merge?
14:51 kahrl BlockMen: go ahead
14:52 kahrl Megaf: if #1512 is so important to you, you might have more luck getting it fixed if you help (e.g. bisecting)
14:58 kahrl it seems you have more luck reproducing that bug than I so it would be great if you did the bisecting
15:00 ImQ009 joined #minetest-dev
15:08 _Esteban left #minetest-dev
15:15 BlockMen pull for named colors https://github.com/minetest/minetest/pull/1720
15:16 VanessaE BlockMen: use unified dyed palette for names ;)
15:17 BlockMen VanessaE, lel
15:17 BlockMen good joke ;D
15:17 VanessaE that'll cover everyone's use cases :P
15:18 VanessaE actually, it's only 12 hues anyways plus light/dark/etc
15:19 _Esteban joined #minetest-dev
15:32 NakedFury joined #minetest-dev
15:36 BlockMen sfan5, did you create this branch? https://github.com/minetest/m​inetest/tree/unknown_node_tex
15:39 hmmmm joined #minetest-dev
15:45 sfan5 BlockMen: yes, accidentally
15:45 sfan5 feel free to remove it
15:45 VanessaE awww
15:46 VanessaE I like that "sick mac" texture :)
15:46 VanessaE seems like a shame to discard it.  can't it be used for something?
15:47 BlockMen VanessaE, if we would have bedrock it could be used
15:48 BlockMen or add an easteregg
15:48 VanessaE well we DO need some kind of bedrock actually
15:49 VanessaE but I don't see how that image would be associated with that
15:49 VanessaE as an easter egg, however, definitely
15:49 VanessaE bedrock has to be done by the core mapgen though, not by Lua.  too slow otherwise
15:50 * nore likes the easter egg idea
15:51 VanessaE just put a few layers of it down at -30924-7, with a slight "landscape" to it (i.e. not perfectly flat), add a couple of .conf settings to control what Y coords it is generated at
15:51 VanessaE simple enough
15:51 VanessaE use ShadowNinja's bedrock node name and textures
15:51 nore <VanessaE> bedrock has to be done by the core mapgen though, not by Lua.  too slow otherwise <-- even with voxel manip?
15:52 VanessaE nore: I'm pretty sure, yeah.
15:53 VanessaE paramat would know better than me, but it's surely faster to write precisely the nodes that need written directly in the engine, than to do a vmanip fetch, fill the nodes in Lua, and write it out
15:53 nore I'm almost sure you can get below 0.1s/chunck
15:53 nore of course, it would be faster in C++
15:53 VanessaE I would think core could do it in a tench of the time, and when it comes to mapgen, it needs to be as fast as possible
15:53 nore but 0.1s/chunck seems already low enough
15:53 VanessaE this is a case where specializing is absolutely recommended.
15:54 Zeno` 60ms (on my machine) just to get the vmanip data
15:54 kilbith joined #minetest-dev
15:54 VanessaE you could always configure it in the map_meta.txt or minetest.conf or something, and alias the mapgen'd nodes like usual, if you wanted it to be something different than bedrock at the bottom of the world
15:54 nore well, it would be under 60ms to modify it
15:55 nore and perhaps another 60 to write
15:55 nore that would make less that 0.2s total
15:55 BlockMen hmm..i dont thing it would be that slow with VM
15:55 BlockMen i can cenerate jungle in 139 ms
15:55 BlockMen then bedrock should be done much faster
15:55 VanessaE that's already well over 0.1s per chunk.  How much time to write 2*6400 nodes?
15:55 Zeno` In C or C++ I could generate the entire chunk in that time. Sure, 60ms doesn't sound a lot but when you have a server with, maybe, 15 players all triggering mapgen it does make a difference
15:56 hmmmm that's a misunderstanding of how the mapgen works.
15:57 Zeno` hmmmm, really?
15:57 BlockMen sfan5, done
15:57 hmmmm the first player to have that block within their view is the one that starts the generation (or at least queues it)
15:57 Zeno` the players are all over the place though
15:57 hmmmm I don't see how a bedrock mod would be "intensive"
15:58 hmmmm you just check the coordinates, if it's a chunk that's on the edge, start generating bedrock
16:00 VanessaE that's fine for getting the generator going, but that doesn't speak for how long it takes to populate the chunk
16:01 VanessaE it'll certainly be faster than the stuff I usually do of course, since you don't have to check the land for anything, just blindly overwrite it
16:01 sfan5 BlockMen: close the pull request too then
16:01 VanessaE but speed is *critical* here
16:01 Megaf_ joined #minetest-dev
16:01 VanessaE this is core mapgen we're talking about
16:01 BlockMen its closed 3 days already
16:01 hmmmm critical at the nether boundaries of the map...?
16:01 BlockMen *or something like that
16:02 hmmmm how much time are people going to spend there for real
16:02 hmmmm bedrock is mod material.
16:02 VanessaE hmmmm: mapgen speed is always critical, and I've had plenty of players go to the boundaries of my maps in the past.
16:02 VanessaE hmmmm: that's fine if it's in a mod.
16:02 VanessaE hmmmm: but as soon as someone proposes putting it into minetest as a default component, it becomes engine material
16:03 Hunterz joined #minetest-dev
16:03 VanessaE hmmmm: remember, one player milling about in new terrain affects *everyone*.
16:05 BlockMen i dont see any reason to add bedrock to minetest
16:05 VanessaE BlockMen: for completeness.
16:06 VanessaE if it's not bedrock then something else
16:06 VanessaE there should not be an empty void at the bottom of the world.
16:06 VanessaE put unbreakable nyan cats there if you want.
16:06 Zeno` that's right! There should be turtles
16:06 VanessaE or unbreakable bright green cobble.  it doesn't matter
16:06 VanessaE whatever it is, it should NOT be an empty void.
16:06 BlockMen it should be obvious that there comes not more at a deep like -31000
16:07 VanessaE BlockMen: so you think a player standing on an empty void looks good?
16:07 VanessaE to the player, it looks like the map is refusing to load.
16:07 kilbith it's currently better to focus on the snow biomes than bedrock, imho
16:07 VanessaE to the player, it looks like a bug.
16:07 BlockMen VanessaE, do you really think a player standing there says: damn, that is shit. i will stip playing mt now
16:07 Zeno` the player expects turtles
16:08 Zeno` BlockMen, actually they do
16:08 VanessaE BlockMen: actually yes, I have had players say almost that ues
16:08 BlockMen lel
16:08 VanessaE yes*
16:08 VanessaE BlockMen: players see that shit and think "wtf?  there's no bottom?"
16:08 VanessaE they don't quit playing, but they get a negative impression of an unpolished game
16:09 Krock updated #1715
16:09 rubenwardy I agree with adding bedrock, or similar. Maybe a massive lava ocean?
16:09 BlockMen and what will they do at the egde of surface?
16:09 BlockMen do you want there also a wall?
16:09 VanessaE BlockMen: that's a totally different matter
16:10 VanessaE the player isn't ***standing*** there
16:10 Miner_48er joined #minetest-dev
16:11 VanessaE BlockMen: actually a short (<50 nodes tall) wall made of stone wouldn't be out of the question.  It would look like a natural formation
16:11 VanessaE but that's a totally different matter than the bottom of the world, where the player can actually *walk on the void*
16:12 VanessaE MC has its farlands for the X/Z limits, which may as well be the same as those short walls
16:12 BlockMen VanessaE, i dont see any problem here
16:12 VanessaE (difference there is, aside from caves being generated in them, is the world is extremely glitchy past those points)
16:12 BlockMen if you look down its gray
16:13 BlockMen that enough for the bottom of world
16:14 hmmmm i can't understand how you people could get so uptight about what happens at the world's boundaries when there are so many more pressing issues to deal with
16:14 hmmmm what the hell is it with the boundary that fascinates players so much
16:15 VanessaE hmmmm: polish.
16:15 hmmmm if you want to bitch about it that much, you can do it yourself too
16:15 hmmmm yechk
16:15 VanessaE hmmmm: same reason we get upset over the "pressing" issues when people then suddenly go off and start fiddling with stuff no one wants all that much at the moment.
16:15 hmmmm it's NOT my priority and will NOT be ever.
16:16 Calinou joined #minetest-dev
16:16 VanessaE hmmmm: no offense, but that ^^^^ that is why.
16:16 VanessaE particularly the "not be ever" part.
16:16 hmmmm then do it yourself
16:17 * VanessaE sighs
16:17 * VanessaE throws the polish in the trash
16:17 VanessaE this stuff clearly isn't needed here.
16:18 hmmmm i don't get it
16:18 hmmmm you bitch and moan but you'll never do a lick of actual work
16:18 hmmmm that is so annoying
16:18 VanessaE wat
16:18 VanessaE O_o
16:18 hmmmm what's so hard about making a bedrock mod
16:18 Zeno` Hang on, that's not fair or correct at all
16:19 VanessaE you clearly don't play, hmmmm.  one already exists.
16:19 VanessaE G*d damn it man.
16:19 VanessaE MODS ARE TOO SLOW FOR THIS.
16:19 hmmmm why can't you just make it, have people use it, and then if it gets really popular and there are demands to move it into the default game, then just do that...
16:19 VanessaE what part of that is so hard to understand?
16:19 hmmmm no, they are not.
16:19 hmmmm mods are just as slow as plantlife.
16:19 VanessaE you haven't tried the latest plantlife either then
16:20 VanessaE 170ms max latency is slow?
16:20 hmmmm this conversation is a giant waste of time
16:20 hmmmm i'm going to go do something productive
16:20 hmmmm less talk, more code
16:21 Zeno` Has minetest reached version 1.0 yet?
16:21 VanessaE ok, let's take a look at something here:  https://github.com/VanessaE
16:21 VanessaE you see that little chart there at the bottom below the repositories list?
16:21 VanessaE don't you DARE sit there and tell me I've been doing nothing.
16:21 VanessaE that just pisses me off.
16:21 Zeno` If not then the people writing the mods, running the servers and the players are your beta testers. You should, really, pay attention.
16:24 Krock Zeno`, it reached version (0410 / 1000)
16:25 hmmmm VanessaE:  I mean for the core game
16:25 Zeno` In that case then everybody is a beta tester and their suggestions and observations should not be so easily ignored
16:25 hmmmm VanessaE:  If you think bedrock is needed in the base game, why the hell not add it
16:25 hmmmm I've offered you commit privs multiple times and you refused every time
16:26 hmmmm so if something pisses you off, and you have a general consensus of approval, you can just fix it yourself rather than BITCH nonstop in the channel
16:26 VanessaE hmmmm: I bitch because it's all I can do to try to get shit to happen around here.  I don't understand enough C++, let alone enough of the engine's codebase, to do much more than read through the pulls from time to time trying to learn it.  And that's after learning C in college and programming in at least a couple of C-like languages since then.
16:26 VanessaE and because I don't understand and can't seem to wrap my head around it, that's why I refuse commit privs.
16:27 hmmmm you've been involved with minetest longer than i have.  "lack of familiarity" is a horrible excuse... it's because you don't even try to learn C++ or the engine
16:27 VanessaE I DO try.
16:27 VanessaE and I fail every damn time I try to read the code.
16:27 hmmmm if you have a question about the code, why not ask here?
16:27 VanessaE so like you, I go off and do more productive things.
16:27 Calinou <VanessaE> MODS ARE TOO SLOW FOR THIS.
16:27 Calinou my bedrock mod uses C++ ore gen
16:27 Calinou should be very fast
16:28 Calinou has deepstone too, good balance.
16:28 Calinou and doesn't generate too shallow
16:28 Zeno` Seriously, hang on. Stop with the ad hominem stuff :( It makes me cry
16:28 hmmmm Calinou, what ore type do you use
16:28 Calinou https://forum.minetest.net/v​iewtopic.php?f=11&amp;t=9231
16:28 Calinou look
16:28 Calinou I think it's scatter
16:28 Calinou but I would like to have a method to generate flat surfaces too
16:28 Calinou this one replaces stone by bedrock in a large area, which is enough for nearly 100 % of cases, you won't ever load the bottom of the world
16:29 Calinou loading the bottom of the world has bad consequences with mob mods and such, so it should be avoided
16:29 Calinou (mobs will duplicate if they fall in void)
16:29 hmmmm well
16:29 hmmmm you can do that
16:29 hmmmm check out ore sheet type
16:29 hmmmm make the scale of the perlin noise like 1 or something, and then make the offset like 2
16:30 hmmmm if you want it to be 100% straight and flat, make the perlin noise scale 0
16:30 Zeno` it's simplex noise
16:30 Calinou here I don't want it to be straight and flat, would be ugly
16:30 Calinou I replace stone with bedrock, not all blocks
16:30 Calinou (but an ore type that could replace all blocks at once would be useful!)
16:31 hmmmm hmmm
16:31 hmmmm ugh
16:31 hmmmm while i'm mucking around with the ore, why not
16:32 hmmmm i just never had demand for ore that replaces any nodes
16:32 hmmmm why do you let the caves cut through your bedrock?
16:32 hmmmm set is_ground_content = false
16:33 Calinou my ore gen only replaces stone
16:33 Calinou it's intended, I don't want fully flat bedrock – boring, ugly, unnatural
16:33 Calinou so caves go through bedrock, gravel and dirt generates in them occasionally, but all caves eventually end
16:33 Calinou no cave is more than ~200 blocks deep
16:33 Calinou so you don't ever reach bottom of world
16:33 * Calinou knows most people will have trouble understanding :(
16:34 hmmmm no, caves can definitely hit the bottom
16:34 hmmmm -31000 is the generation cutoff, and 80 is the chunk size
16:34 hmmmm so if you have whatever - 80 <= -31000, it'll cut off generating that chunk
16:36 hmmmm if i recall, the central chunk at the origin is offset in the negative axis by 2 mapblocks
16:36 hmmmm so that means it's -[(387 - 1) * 80 + 32]
16:36 hmmmm = -30912 when the apparent map should cut off
16:37 Calinou <hmmmm> no, caves can definitely hit the bottom
16:38 Calinou install bedrock mod, try to find cave which goes from above-bedrock to void
16:38 Calinou probably will take hours, by doing it intentionally
16:38 hmmmm just because it's uncommon doesn't mean it won't happen.
16:38 Calinou yeah, I know, “IPv4 is probably enough” :P
16:39 hmmmm in any case, I'll agree that the boundary of the world must be better defined
16:39 Calinou we could extend world size by increasing mapgen limit
16:39 hmmmm 2^15 - 31000 = 32768 - 31000 = 1768 of buffer zone
16:39 Calinou it crashes if you go too close to 32768, but you can still add some blocks?
16:40 Calinou like, ~31500
16:40 hmmmm things wouldn't crash
16:40 hmmmm but they'd definitely screw up
16:41 hmmmm like it'd insert a duplicate (x, z) for a completely different MapSector
16:41 hmmmm and again, that's only if we remove the safety checks that are already there
16:42 hmmmm what I'm proposing is that 31000 and -31000 are the actual boundaries, i.e. that's where map stops generating
16:42 hmmmm or maybe i guess we could bump it to +-32000
16:43 hmmmm so if you have a chunk starting at 31980, for example (i know it's not a valid chunk start coordinate), it'll generate map for 31980 to 32059
16:43 hmmmm and then on blit back, it'll cut off anything after 32000
16:43 hmmmm so in effect only 20 nodes will be written
16:44 hmmmm that's 16 in the first mapblock of the chunk, and 4 in the next mapblock, followed by the rest as content ignore
16:54 kaeza joined #minetest-dev
17:02 casimir joined #minetest-dev
17:10 kaeza joined #minetest-dev
17:16 cg72 joined #minetest-dev
17:20 Krock Who of the devs can give me thumbs up for #1715 ?
17:29 Hunterz joined #minetest-dev
17:34 _Esteban joined #minetest-dev
17:52 ShadowNinja VanessaE: My bedrock mod's very fast and short-circuits if it isn't generating a chunk with bedrock.  Bedrock definitely doesn't need any extra core support.  It's generation is very simple so almost all of the time will be spent reading and writing the data.
17:55 VanessaE left #minetest-dev
17:55 Hunterz1 joined #minetest-dev
17:56 jin_xi so i have updated particles pull, if you tried it before its working better now
17:58 alexxs joined #minetest-dev
18:25 kaeza_ joined #minetest-dev
18:31 Calinou_ joined #minetest-dev
18:34 nore joined #minetest-dev
18:48 shadowzone joined #minetest-dev
18:48 shadowzone joined #minetest-dev
19:08 kilbith joined #minetest-dev
19:13 zat joined #minetest-dev
19:26 VargaD joined #minetest-dev
19:29 kilbith joined #minetest-dev
19:36 Hunterz1 left #minetest-dev
20:00 shadowzone left #minetest-dev
20:05 elinvention joined #minetest-dev
20:23 Amaz joined #minetest-dev
20:52 kilbith joined #minetest-dev
22:03 kaeza_ joined #minetest-dev
22:37 proller joined #minetest-dev
22:53 proller joined #minetest-dev

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