Time  Nick      Message
13:59 sfan5     will merge #16735 later if nobody complains
13:59 ShadowBot https://github.com/luanti-org/luanti/issues/16735 -- Document integer data types and ranges in Lua API by Wuzzy2
15:23 Krock     quickly browsing through..
15:28 Krock     > Integer range of depth: [0, 32767]
15:29 Krock     The notation [0, 1234]  vs [0.0, 1234.0]  could be used to differentiate between floats and integers. Anyway. this is good too.
16:27 cx384     Oops there is now a empty file named "2" tracked by git. I probably had a typo in my bash script. (I shouldn't work on PRs late at night anymore.)
16:28 sfan5     easter egg
16:29 cx384     xd Will push https://github.com/cx384/minetest/commit/4279e2f7a59939f68506fbc899d507d54093ef47 to master in 15 min. (probably better than a force push since it is already 19 hours old)
16:31 Krock     sure
16:34 sfan5     alternative I can include the removal in my misc PR
16:34 sfan5     alternatively*
16:38 cx384     I already made the commit, not need for extra work.
16:38 cx384     * no
16:43 MTDiscord <rollerozxa> at least you didn't accidentally leak all your env variables into the commit message :) https://github.com/apache/httpd/commit/3ad6cba4b484b166fd3b1c6272e383052bc95624
16:44 MTDiscord <exe_virus> Looking at the roadmap, and current state of things, looks like we started down the path of SSCSM (woo), also looks like UI and input is being tackled.   Only thing I'm surprised is not included yet is a camera API, either SSCSM or otherwise. I think it's already been started, but was hoping for extensions such as roll, besides just pitch and yaw. Will take a look at GitHub to see if anything has been worked towards roll, but just that
16:44 MTDiscord alone provides a ton of flexibility
16:45 cx384     The problem was that the file is empty and git diff didn't show much. If it weren't empty I would have noticed it.
16:48 cx384     pushed
16:49 MTDiscord <exe_virus> Okay, looks like the concept it approved, just not yet implemented, there's a now closed PR. maybe I can tackle this one
19:06 sfan5     does anyone know why we violate lua behavior with our protected metatables for userdata objects?
19:07 sfan5     from lua getmetatable(VoxelManip()) is something like { get_data = <function>, ... }, but this wouldn't normally ever work
19:07 sfan5     it works because the hidden metatable then has an __index that points to the methods
19:07 sfan5     the expected normal lua behavior would be getmetatable(VoxelManip()) == { __index = { get_data = <function>, ... } }
19:13 sofar     sfan5: https://gist.githubusercontent.com/sofar/360ba46b02b6fc3be0af87a2f0547084/raw/24058c78e261d74cd130e32f30235d0ac5bbbf2d/gistfile1.txt
19:13 sofar     perhaps it's utter BS.
19:15 sofar     The fix would be to either:
19:15 sofar     - Set __metatable to a dummy value (e.g. true or "protected") so getmetatable() returns nothing useful
19:15 sofar     - Or make the methodtable read-only (by giving it its own __newindex that throws an error)
19:17 sfan5     it's BS in the sense that it missed the point - twice
19:17 sfan5     I already know how we made it this way because I looked at registerClass myself
19:17 sfan5     and it's not about "what happens if lua writes to the method table" at all
19:19 sofar     oh, I misread your first sentence entirely
19:20 sofar     I guess it's taking a pickaxe to git for some archaeology
19:21 sfan5     likely result: it was like this in the first commit with no indication why
19:21 sofar     d6ac3d8d9a
19:22 sofar     "bingo"
19:22 sofar     actually the commit msg indicates intent, so, it's half a bingo?
21:02 sfan5     @luatic what are you thoughts on including something like this with the engine's dump()? https://github.com/sfan5/luanti-developerconsole/blob/master/dump.lua#L14
22:16 MTDiscord <luatic> sfan5: would be neat to get more compact output, the checks there also seem reasonable, though that's always a matter of some finicky tweaking and taste
22:17 MTDiscord <luatic> when i wrote a lua formatter in js some years back i had a condition that consisted of number of fields and serialized length of each field to decide whether to keep tables inline