Time Nick Message 00:40 lhofhansl re: https://github.com/minetest/minetest/blob/master/src/server/clientiface.cpp#L88 00:43 lhofhansl I added that for player position prediction when loading blocks. Only LuaEntitySAO has a getVelocity method. Of the object is not a LuaEntitySAO, the player's speed will be used instead - in worst case the position is not as accurate as it could be. 01:52 MTDiscord Re: testing. Yeah I've been thinking through how I would go about testing this and a LOT of other things, I think it'll have to be something where I make a test suite based on dev test, where I can test performance, interaction issues, etc. 02:24 MTDiscord Should setting is_visible = false work on players? 04:29 MTDiscord Should Yes, does 🤷‍♀️ 04:44 Mantar I had it fail to work on players before 04:45 Mantar Instead I set the model size to 0,0,0; also replaced the nametag with " ", which hides it 04:53 MTDiscord Yeah, it doesn't work for me at all with latest dev + observers PR. I have to set visual_size to accomplish what I want, but it looks like I do that in code from a few years ago too though, so I suspect it might just be missing doc 10:00 Desour re reliability of lbms: I believe there were issues where if you insert partial mapchunks, e.g. via worldedit, run_at_every_load=false lbms won't get executed for these blocks. I don't have evidence for this claim though 10:37 sfan5 hmm yeah that could be 10:37 sfan5 however at the time of insertion the mapblock just contains ignore 20:18 MTDiscord Tbh we've had LBMs failing in weird situations. For example, rails got a bit of a rework in VL (not merged yet) and old rail nodes, which are defined somewhat differently than the new system, would have an LBM to convert them into new ones. Some of the people testing the PR found the old nodes ("missing texture") unconverted. We haven't managed to track down why it failed, but it sure was in some generated structures that already had 20:18 MTDiscord been generated, in an old world.