Time Nick Message 02:09 MTDiscord convert to STRINGS, not tables. 02:10 MTDiscord then its just a == 02:11 MTDiscord actually, is the order of meta entries in itemstrings defined? 02:11 MTDiscord how does to_string decide the order to put them in 06:33 MTDiscord no, the order is not define; converting to string does not work 06:33 MTDiscord ItemStackMetaData uses typedef std::unordered_map StringMap;, which does hashing in the end 11:18 MinetestBot 02[git] 04SmallJoker -> 03minetest/minetest: Content tab: Fix content download broken by bc3dccc 13b5e7280 https://github.com/minetest/minetest/commit/b5e7280708221cdcca89df7d16f2aea19e4b3c4c (152022-09-16T11:01:23Z) 11:20 MinetestBot 02[git] 04Wuzzy2 -> 03minetest/minetest: Add paramtype2s for 4 horizontal rotations and 64 colors (#11431) 131d04903 https://github.com/minetest/minetest/commit/1d04903c1973591e795d3275b900d76d7cb7877a (152022-09-16T11:18:55Z) 11:20 MinetestBot 02[git] 04vilhelmgray -> 03minetest/minetest: Add support for MINETEST_USERDATA environment variable (#12639) 1319e9363 https://github.com/minetest/minetest/commit/19e936362aab3806a9554ab811be0562dcb41509 (152022-09-16T11:19:44Z) 11:20 MinetestBot 02[git] 04savilli -> 03minetest/minetest: Fix UAF in craft recipes (#12763) 139428917 https://github.com/minetest/minetest/commit/942891787028d0cbce226c2af0536a18d5a37df2 (152022-09-16T11:20:14Z) 16:01 Zemtzov7 Hello there, I see `[ConnectionSend] con(4/1) Packet quota used up for peer_id=#, was # pkts` warnings in the server logs pretty often. What it means and is it fixable anything other way than tweaking the `connectiontreads.cpp`? (MT 5.6.0) 16:18 muurkha I had that problem a lot, associated with desynchronization "lag" that sometimes made the game unplayable; other people here have said that turning on the WAL in the server's SQLite configuration, or switching it to Postgres, solved the problem for them 16:18 muurkha I haven't had it lately 16:19 sfan5 does it appear regularly or once in a while? 16:22 MTDiscord The SQLite WAL or switch to Postgres thing seems to be a big improvement on systems with a disk I/O bottleneck, such as mechanical drives or SD cards. I haven't seen evidence that it has a big impact on other bottlenecks such as compute, but would be curious if anyone has. 16:22 muurkha it shouldn't affect other bottlenecks such as compute 17:05 Zemtzov7 sfan5: mainly regularly, depends on players. [if you asked it about my message] 17:07 sfan5 I was 17:07 sfan5 if the message appears often & repeatedly then you have a performance problem 17:08 Zemtzov7 I counted 1969 warnings for 1 day... 17:08 sfan5 but most likely not in the sense of "you need to throw more hardware at it", rather "a mod is doing something stupid and should be fixed" 17:08 MTDiscord was that day related to the moon landing in one way or another? 17:09 Zemtzov7 depends on `max_lag` too? 17:11 Zemtzov7 now its 1.45s with 6 players. Yeah we have some mods from APercy installed(blimp,ju52,pa28), probably they mostly causing lags 18:28 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Bump used IrrlichtMt version 13a428a0c https://github.com/minetest/minetest/commit/a428a0cf37581a35f9c4f81c2e71633e6cc3dbb9 (152022-09-16T17:54:59Z) 20:03 farribeiro I consider that luajit as a dependency is very old, the last release was in 2017 20:04 MTDiscord most people use the latest 2.1 dev version 20:05 farribeiro could be in touch with the luajit team for a new beta release or promote the stable 20:06 MTDiscord lol it's not like people haven't already tried 20:06 MTDiscord the advice from upstream luajit is to run latest commit because version numbers are obsolete 20:06 muurkha heh 20:07 farribeiro theses days i suggest Calinou use the latest commit of luajit 20:07 farribeiro from git 20:08 farribeiro Calinou maintains the flatpak minetest... https://github.com/flathub/net.minetest.Minetest/pull/43 20:09 farribeiro i try out too use lua 5.4... worked +- 20:09 MTDiscord please don't use newer PUC lua 20:10 farribeiro i see... but i tested, works without big problems 20:10 MTDiscord No. 20:10 MTDiscord It's not backwards compatible and the errors are likely to be subtle. 20:10 farribeiro but the contentdb fails 20:10 MTDiscord ? 20:10 MTDiscord Particularly because of the introduction of an integer type. 20:11 MTDiscord if contentdb fails that's because it's not supported. 20:11 MTDiscord Bumping Lua to 5.4 would require a major version update and dropping LuaJIT support. 20:11 MTDiscord (oh, and fixing all of our Lua codebase and everything that interacts with it and the in the end we'll still have missed some subtle bugs) 20:13 muurkha Lua is not really intended for people to update to new versions 20:13 muurkha the idea is that you copy some version of Lua into your game or whatever and then just keep using that version of Lua forever 20:13 MTDiscord Where do you get that from? 20:13 MTDiscord I mean that's what everyone's been doing since 5.1 is pretty decent 20:13 muurkha something or other Ierusalimschy wrote years ago 20:14 muurkha unfortunately I read it many years ago so I don't have a link handy 20:14 muurkha but he wants to have the freedom to make backward-incompatible changes 20:14 MTDiscord We don't need to drop jit for 5.4. Openresty has a 5.4 jit fork 20:15 Desour I've heard openresty is bad 20:15 MTDiscord I guess then "bump lua version" may be put on the breaking changes list 20:15 MTDiscord The reputation of openresty is rather irrelevant. As long as their jit fork works 20:15 muurkha yeah, "bump lua version" is more like a "break everything" change 20:16 muurkha it's not like bumping a Python version or a Perl version or a GCC version 20:16 Desour just a recent quote from mikepall: > Stop reporting OpenResty bugs here! It's many commits behind, it 20:16 Desour has lots of bad/broken patches and even broken architectures. 20:16 MTDiscord TBF I don't like the addition of integers in 5.3 20:16 Desour src: 20:17 MTDiscord Is the openresty JIT backwards-compatible with all the FII-dependent libraries out there? If not it's a hard no from me. 20:17 MTDiscord "Many commits behind" isn't applicable, the codebase is substantially different 20:17 Desour anyway, why should we increase the lua version and break all mods? 20:17 MTDiscord because change, goddamnit! 20:17 MTDiscord /s 20:18 muurkha it might be nice to import some of the features from newer Lua versions 20:18 MTDiscord I'd gladly sacrifice the mod ecosystem to get first class bitops 20:18 MTDiscord :] 20:18 muurkha I was thinking about those, yeah 20:18 MTDiscord bitops aren't nearly as useful as you make them out to be 20:18 MTDiscord For you 20:18 muurkha but that's not in itself a backward incompatible change 20:18 MTDiscord you can emulate 99% using floor, mod and pow, sub and div 20:18 muurkha you can just use the bitops module, you don't have to do that 20:19 MTDiscord They are infinitely useful in every application you haven't needed, apparently 20:19 MTDiscord and for the remaining stuff we have the bit library now which operates on 32 of the 64 double bits 20:19 Desour you don't need to emulate, we have the bitop lib, luatic 20:19 MTDiscord Desour: I'm aware 20:19 MTDiscord GreenXenith: I wrote a PNG reader using my emulated bitops. It's definitely possible. 20:19 muurkha for most code I'd agree that bit ops aren't very useful, but packing stuff bitwise into node params is pretty common for mods 20:19 MTDiscord Possible doesn't mean good 20:20 MTDiscord muurkha: That is an example where mod, div-sub / floor work exceptionally well. 20:20 Desour luajit has some extensions from newer versions, btw.: https://repo.or.cz/luajit-2.0.git/blob_plain/refs/heads/v2.1:/doc/extensions.html 20:20 MTDiscord Besides, ther are plenty more features than bitops 20:20 MTDiscord There* 20:20 muurkha GreenXenith: what are your favorites? 20:21 MTDiscord The problem stems in part from the fact that due to double representation, bitops on a double with an uint value won't necessarily work the way you might think they work 20:21 MTDiscord but if you use emulation, everything is fine, and your hardly need more Lua VM ops 20:21 MTDiscord Constants might be nice, and there's some newer table methods/metamethods that I'd like 20:22 MTDiscord Also closeables :] 20:25 muurkha which metamethods? 20:27 MTDiscord Can't recall at the moment 20:27 MTDiscord well, you get the entire bitops as syntactic sugar 20:27 muurkha luatic: closeables are finalizers? 20:27 MTDiscord also __len works I believe? 20:28 MTDiscord muurkha: you can mark local vars as closeable / "to close" to have them be closed when they go out of scope 20:28 muurkha I guess that's not quite the same thing as finalizers 20:28 muurkha maybe a better alternative 20:29 MTDiscord I suppose you would call the __close handling metamethod the "finalizer"? 20:30 muurkha sort of, but going out of scope isn't the same thing as being garbage collected 20:30 MTDiscord well, Lua has __gc for that 20:31 muurkha oh I didn't remember that 20:31 muurkha the bitop syntactic sugar can be nice for embedded DSLs. parser combinators, that kind of thing 20:48 * Desour likes how the remover testtool in devtest looks like a cross. it makes it look like you're doing some sort of exorcism