Time  Nick       Message
12:03 sfan5      merging #14238 in 10m
12:03 ShadowBot  https://github.com/minetest/minetest/issues/14238 -- [no squash] Unittest and some documentation things by sfan5
15:22 [MTMatrix] <Zughy> has any core dev thought about asking sfence if they want to become one, provided they've got what it takes? Their contributions amount in these last two months has been insane https://github.com/minetest/minetest/pulls/sfence
15:24 Krock      they may ask at any time
15:24 Krock      and we could look at it latest in the next meeting
15:27 Krock      will push https://pastebin.com/raw/KCYsn3yz in 20 minutes to fix #14241
15:27 ShadowBot  https://github.com/minetest/minetest/issues/14241 -- Arithmetic exception with [crack texture modifier
15:48 Krock      pushin
16:36 MTDiscord  <luatic> merging #14042 in 10m
16:36 ShadowBot  https://github.com/minetest/minetest/issues/14042 -- Add function to get all hud elements by cx384
22:15 sfan5      hmm what are those spurious CI failures
22:26 MTDiscord  <luatic> Not sure. The CI ran fine on that PR. The timeout message it fails with doesn't make much sense.
22:33 sfan5      the server is supposed to shut down after the tests https://github.com/minetest/minetest/blob/92c55c27cf520bd7b0e59639cc7cce7bcaf5d747/util/helper_mod/init.lua#L14
22:38 sfan5      or are the tests simply taking too long?
22:39 lhofhansl  Hi... I've been looking at MapBlock::getDayNightDiff(). We use that extensively to decide whether we should send blocks to the client or not. The problem is that solid blocks would also not be send, so if they happen to be at the surface we see holes. Hence we have (1) do not use optimization underground (y < water level), or (2) the block is further away than block_send_optimize_distance.
22:39 lhofhansl  I just realize the main benefit of this optimization is not send all-air blocks. So I'd like to change that to that.
22:40 lhofhansl  This bit is unfortunately serialized, so I'd need to increase the serialization version and deal with this accordingly in new and old servers.
22:41 lhofhansl  Any comments/concerns? The benefits: (1) no more holes, (2) No need to check for underground, (3) no block_send_optimize_distance, we can always use this optimization.
22:43 sfan5      can you put this into an issue? then I can look at it tomorrow
22:46 lhofhansl  Will do. Risk: Some underground solid blocks might be sent if occlusion culling did not catch them, and cost of occlusion culling might go up slightly.
22:47 lhofhansl  I can make PR (without the version handling) to try it out, too. Would have regenerate the world as the meaning of the day/night bit would change.
22:47 sfan5      speaking of issues I have bad news for one of your commits ;) #14261
22:47 ShadowBot  https://github.com/minetest/minetest/issues/14261 -- `test_mapgen_edges` test runtime regression
23:12 lhofhansl  Hmm. Ok. Pretty sure the code is correct, so likely the test code needs to be adjusted.
23:13 lhofhansl  Or there's some weird thing with generating.
23:14 lhofhansl  Ohh... Past the edges. Maybe we only check that in RemoteClient, but not from the active object code.
23:14 lhofhansl  Need to head out for a few hours, will have look later today or tomorrow.