Time  Nick       Message
13:36 rubenwardy I'll tag 5.1.1 tonight unless there are any objections
13:37 rubenwardy Actually, should I pull translations in if the template hasn't been updated since 5.1.0?
13:37 rubenwardy Would keep Hamlet happy
14:12 sfan5      can we merge translations "twice" safely?
14:13 sfan5      as in, pull it into backport-5 and master
14:20 Wuzzy      https://github.com/minetest/minetest/pull/9294
14:21 p_gimeno   sfan5: I guess you can cherry-pick or rebase
14:21 Wuzzy      !title
14:22 ShadowBot  Add callbacks for falling and attached nodes by Wuzzy2 · Pull Request #9294 · minetest/minetest · GitHub
15:02 p_gimeno   re https://github.com/minetest/minetest/pull/9205#issuecomment-571717643 : the dir does not have enough information, it lacks roll info
15:04 p_gimeno   at least for facedir
15:07 p_gimeno   the rhotator mod includes code that builds a matrix for each facedir, to use as a LUT, see https://github.com/entuland/rhotator/blob/master/init.lua#L116
15:08 p_gimeno   actually https://github.com/entuland/rhotator/blob/master/init.lua#L63
15:10 p_gimeno   with that, you can use a matrix to euler conversion code
15:10 p_gimeno   that will give you a LUT for facedir-to-euler
15:12 p_gimeno   I vouch for the exactness of the rhotator formulas
15:14 p_gimeno   for matrix to euler code, see https://github.com/minetest/minetest/blob/master/src/util/numeric.cpp#L199
15:34 rubenwardy sfan5: surely git is clever enough for that
15:34 rubenwardy It's the same commits
15:57 p_gimeno   Wuzzy: http://www.formauri.es/personal/pgimeno/pastes/facedir2euler.lua (re #9205)
15:57 ShadowBot  https://github.com/minetest/minetest/issues/9205 -- Falling nodes: Add support for facedir, colorfacedirr, wallmounted, colorwallmounted, color, airlike, signlike, torchlike and glow by Wuzzy2
15:58 p_gimeno   wait, there's a problem with element 0, you need to use facedir_to_euler[facedir + 1] because it's 1-based
16:00 Wuzzy      this might be useful for a Lua API function
16:02 p_gimeno   it needs testing, because the matrix might be transposed when converting to euler, so please test with a node where you can distinguish rotation easily and which has several different rotations
16:02 Wuzzy      i currently try the math.atan2 solution
16:03 Wuzzy      so you need to test yourself atm, sorry. You can use [devtest]. this contains a facedir test node ?
16:05 p_gimeno   the math.atan2 solution simply can't work
16:06 Wuzzy      not if you also take the face rotation into account
16:06 Wuzzy      I have already made it work for wallmounted
16:07 p_gimeno   ah ok
16:07 p_gimeno   I meant as written by SmallJoker
16:07 Wuzzy      yeah Krock's solution does not work
16:09 p_gimeno   wallmounted is easier because you only have 6 directions with no roll (IIRC)
16:13 p_gimeno   there are multiple eulers for the same rotation for 8 of the facedirs, so don't be surprised if your result doesn't match mine
16:13 p_gimeno   (the other 16 should match)
16:22 Wuzzy      hrm
16:22 Wuzzy      i just noticed facedir_to_dir only gives me 1 of 4 possible directions
16:23 Wuzzy      oops
16:23 Wuzzy      nvm
16:30 p_gimeno   facedir_to_dir does not give you the face rotation, it just returns one of 6 possible directions
16:31 Wuzzy      yeah i know. i have to extract the face by hand
16:31 Wuzzy      but even that is not enough, actually
16:31 p_gimeno   it should
16:31 Wuzzy      sigh
16:32 Wuzzy      I am starting to wonder why we're even doing this
16:32 Wuzzy      I am about to give up
16:36 p_gimeno   working with Minetest's axis convention is very tricky
16:37 p_gimeno   most formulas you find around don't work
16:40 Wuzzy      I think this calls for more library functions
16:41 p_gimeno   well, set_rotation is a very recent addition; it's kind of logical to add facedir_to_euler to the set, so that a facedir can be used in set_rotation
16:44 p_gimeno   I have forgotten a lot of things. I don't remember what facedir_to_dir returns, but I believe it's the Y vector. And IIRC, roll rotates over a different vector, which means you can't apply roll over the vector returned by facedir_to_dir.
16:45 p_gimeno   Very tricky, as I said.
16:46 Wuzzy      p_gimeno: i just tested your facedir_to_euler. it does not work, sadly
16:46 p_gimeno   ok, the matrix is probably transposed, let me fix it
16:47 p_gimeno   did you remember to use facedir_to_dir[facedir + 1]?
16:47 p_gimeno   err facedir_to_euler[facedir + 1]
16:47 Wuzzy      yes
16:48 p_gimeno   http://www.formauri.es/personal/pgimeno/pastes/facedir2euler_2.lua
16:49 p_gimeno   reload if you have opened the link
16:49 Wuzzy      nope
16:50 p_gimeno   nope you haven't opened the link or nope it does not work? :)
16:51 Wuzzy      its still wrong but its better than before
16:51 Wuzzy      facedir=0-3 are correct
16:51 Wuzzy      then facedir 4, 8, 12, 16 and 20 are correct
16:51 Wuzzy      the rest is wrong
16:53 p_gimeno   let me check the getPitchYawRollRad function more carefully to see if I screwed it up
16:54 Wuzzy      i strongly recommend to use [devtest]. you can test the facedir stuff much faster
16:55 Wuzzy      oh wait you still need to edit builtin...
16:55 p_gimeno   the function is right, that can't be it
16:56 Wuzzy      p_gimeno: look in my PR to see the correct values
16:57 Wuzzy      basically i just got all the values by trial-and-error... ?
16:59 p_gimeno   why not make a table with that then?
17:00 Wuzzy      heh
17:00 Wuzzy      well it would be nice to also have the formula that produced the table...
17:01 Wuzzy      I am not sure what exactly is needed to make Krock happy anyway... lol
17:02 p_gimeno   I'm not sure what the differences are with my version, I think it should match, maybe one sign convention is inverted
17:04 p_gimeno   ohh
17:05 p_gimeno   you have angles in range 0 to 2pi
17:05 p_gimeno   I have angles in range -pi to pi
17:05 p_gimeno   I think that could explain the differences
17:05 p_gimeno   an angle of 7.8539816339744828 is very strange :)
17:07 p_gimeno   http://www.formauri.es/personal/pgimeno/pastes/facedir2euler_3.lua <-- uses your empirical results
17:13 Wuzzy      p_gimeno: my results, converted to lua: https://pastebin.com/raw/ihS5qAqf
17:14 p_gimeno   let me see if I can output in that format
17:15 Wuzzy      i just searched-and-replaced the constants by hand
17:19 p_gimeno   http://www.formauri.es/personal/pgimeno/pastes/facedir2euler_5.lua
17:20 Wuzzy      ah
17:20 Wuzzy      i see a mistake already, its the last one
17:21 p_gimeno   these could be equivalent
17:22 p_gimeno   there must be at least 16 equal to each other to be considered good
17:23 Wuzzy      wait, what?
17:23 * Wuzzy    test
17:23 p_gimeno   there are several eulers that map to the same rotation
17:23 Wuzzy      ah right
17:24 Wuzzy      but its wrong ? i tested
17:24 p_gimeno   ok
17:24 Wuzzy      my table works
17:24 Wuzzy      actually
17:24 Wuzzy      its the same problem as before
17:24 Wuzzy      fdir 0-3, 4, 8, 12, 16, 20 work, all other dont
17:25 Wuzzy      oh wait. 22 also works. interesting
17:25 p_gimeno   it's sounding more and more like a sign problem
17:26 p_gimeno   still, the problem was getting the table, now you have a table
17:27 p_gimeno   I get dizzy working with MT's axis conventions
17:28 Wuzzy      lol
17:30 p_gimeno   really, with normal conventions this would be a breeze, but this way you have to rework every formula from the beginning in order to not screw up axes or signs, that's what caused #7927
17:30 ShadowBot  https://github.com/minetest/minetest/issues/7927 -- get/set_rotation and get/set_yaw problems
17:30 p_gimeno   I haven't done that, and as a result the formulas don't work
18:08 Wuzzy      hey Krock you have missed a long and productive discussion with p_gimeno about the the rotation in falling_fixes PR
18:10 Wuzzy      p_gimeno experimented with generating the  facedirs a little http://www.formauri.es/personal/pgimeno/pastes/facedir2euler_5.lua
18:11 Krock      calm down. post is still in progress
18:11 Wuzzy      LOL
18:11 Wuzzy      sorry
19:51 Wuzzy      lol Krock your prediction came almost true. i did file a bug for dropped items, but for a completely different reason.
19:54 Krock      in 2020 we'll reach 1000 bugs
19:54 Krock      help to reach this goal
19:58 mmuller    haha, I presume you mean by filing bugs rather than adding them to the code?
20:00 Krock      mmuller: one or another will cause that as final result
21:26 paramat    merging #9265
21:26 ShadowBot  https://github.com/minetest/minetest/issues/9265 -- Improve Lua API documentation on sounds by Wuzzy2