Time  Nick      Message
08:59 celeron55 https://github.com/minetest/minetest/pull/7924
09:00 celeron55 i guess this should go in soon
09:03 celeron55 if there's a problem with some specific mod that can be solved via an issue
09:07 sfan5     I hope it introduces no bugs but at this point it's better to just merge it and then fix stuff than letting it sit
09:08 celeron55 what's the deal with the "VS 2019 x86-portable" build failing?
09:09 celeron55 well, clearly not related to anything the PR does
09:09 sfan5     VS should have been fixed by either #10256 or #10089
09:09 ShadowBot https://github.com/minetest/minetest/issues/10256 -- Fix build for Visual Studio (explicitly cast pointers) by SeekingMeaning
09:09 ShadowBot https://github.com/minetest/minetest/issues/10089 -- Disable pgsql in VS 2019 workflow by adrido
09:15 celeron55 i guess technically ClobberXD was a core dev at a point in time he also had written "tested, works" in the PR so it's one agreement 8)
09:17 celeron55 sfan5: i'll merge it if you plop a +1 in the comments
09:18 celeron55 oh there's SmallJoker already, whatev
09:18 celeron55 +er
12:02 Krock     70us to parse a long (~600 chars) ItemStack string can result in 100ms per MapBlock when chests are filled
12:03 Krock     Additionally, the mapblock serialization structure looks horrible..
12:04 sfan5     what do you mean by that?
12:13 Krock     kinda-unused serializeNetworkSpecific functions, ancient code leftovers for 8 year old maps, old mapblock versions are not upgraded automatically, ItemStack metadata is serialized to JSON by guessing whether it's needed (just do it always?)
12:13 Krock     and in between there are quite a few i/ostringstream copies which should be possible to work around
12:15 Krock     guess I'll do some quality changes prior thinking about performance optimizations
12:17 Krock     and for whatever reason the setting "ignore_world_load_errors" exists - it's obscure to me
12:20 sfan5     serializeNetworkSpecific can definitely be removed, compatibility for pre-0.4 maps could probably be removed too
12:25 Krock     question is how server owners would upgrade those old blocks. if a block in air isn't touched for maaany years, it's still in the old format
12:26 Krock     and migration does not convert the blocks
13:16 celeron55 move the legacy support to a "convert mapblock to a reasonably new version" function?
13:17 celeron55 there's no real need to remove support, just tuck it away in a bunch of code that won't get in the way of new things
13:17 Krock     that already happens for very old versions. for newer ones, it would result in much duplicated code
13:18 celeron55 there needs to be a way to convert all of the world though
13:20 Krock     basically requires launching the server as a whole
13:20 Krock     perhaps as a command like the world-wide /clearobjects
14:18 Krock     does anyone have an old (but small) map somewhere?
14:22 Krock     hmm okay. found one
14:22 sfan5     you can always generate a new one
14:23 sfan5     (with a windows build in wine, compiling old versions can be quite the effort)
14:28 Krock     right. I now found some old worlds in the "Maps" forum. Those should still work in 5.x judging by the compat code
14:57 Krock     very cool ServerEnvironment. Accesses to EmergeManager after it is deleted in Server's destructor
15:04 Krock     2020-08-15 17:03:52: WARNING[Main]: ServerEnvironment::deactivateFarObjects(): id=11 m_static_exists=true but static data doesn't actually exist in (-8,1,11)
15:04 Krock     no crash but this monstrosity
15:11 Krock     #10293
15:11 ShadowBot https://github.com/minetest/minetest/issues/10293 -- Add /upgrade-mapblocks command by SmallJoker