Time Nick Message 01:40 fluxionary "how lua works"? 01:46 fluxionary another thing. i've got my fps_max_unfocused setting set to 5. this seems to sometimes result w/ my client getting massively desynced w/ the server i'm connected to. if i leave for a couple hours and come back, it will then replay the events from the server at 5x or 10x the speed, over several minutes, while it catches up 01:47 fluxionary is that an issue that's been reported already? i've searched before but couldn't find anything 01:48 fluxionary i *think* it only happens if i'm near a mapblock that's changing a lot (e.g. running mesecons mechanisms) 01:48 MTDiscord off the top of my head, i havent seen an issue for that 01:49 MTDiscord if there is one, it will be zughy or I closing your duplicate anyways 01:49 fluxionary i haven't reported it cuz i'm not 100% on a replication, the few times i've tried in single-player, i haven't been able to 02:39 muurkha how Lua works is that you integrate some version of Lua, like 5.0, into your project, and keep using it forever 02:39 muurkha because newer versions of Lua are never backward-compatible 03:16 fluxionary the changes w/ `LUAJIT_ENABLE_LUA52COMPAT` aren't backwards compatible, but they also aren't hard to modify code to use, and also not likely to cause problems for most minetest mods 03:17 fluxionary the most successful software brands break backwards compatibility regularly and still continue to get used 03:22 Blockhead256[m] the most successful software brands have enough resources to keep up with churn and enough of a 300-pound gorilla status to make anyone who complains move on anyway. I suspect we have neither of those 03:28 muurkha fluxionary: the most successful software brands that use Lua don't do that. WoW is still using 5.0 or whatever they settled on when they launched the game 03:30 muurkha Microsoft Windows just recently dropped compatibility with Win16 software from 01989 but still supports Win32 even though you can't buy a new 32-bit CPU to run Microsoft Windows on since like 02009 03:30 muurkha Win32 is from like 01994 03:31 muurkha Linux and MacOS meticulously maintain their compatibility with VT-100 terminals from 01976 03:34 muurkha Linux distributions are still mostly compatible with X-Windows from 01987, although the switch to Wayland is controversial because although it's compatible with regular X-Windows applications, it's not compatible with X-Windows window managers 03:35 fluxionary yeah, but you can't run new software on 1990s-era machines and operating systems. what? 03:35 muurkha Java will still run bytecode compiled in 01996 03:35 muurkha no, you can't 03:35 Blockhead256[m] muurkha: by the way, what's with the 0-prefix on your dates? Is it some sort of "long now" idea? 03:35 muurkha yeah 03:36 fluxionary i guess that point still stands 03:36 Blockhead256[m] anyway flux are you volunteering to fix everyone's mods? 03:37 fluxionary Blockhead256[m], possibly, depending on the scope 03:37 fluxionary Blockhead256[m], my assumption is that the actual changes won't break more than a couple mods 03:41 fluxionary Blockhead256[m], you mentioned advtrains, which i know but am not intimate w/, and dbf, which i've never heard of and isn't in the content db, nor can i find via web search 03:42 Blockhead256[m] dbf was a typo of dbg, the technical name of LMD's debug mod 03:42 Blockhead256[m] https://forum.minetest.net/viewtopic.php?f=9&t=28372 03:42 fluxionary potentially a lot of problems could be detected w/ the proper automated testing framework 03:43 fluxionary oh dbg looks like something i want to get to know better, thanks for the link :) 03:44 Blockhead256[m] just for the pretty print and better lua command than worldedit, it's pretty handy. The stepping is ok too, though I usually find it slower to work with than print statements (I know...) 03:46 fluxionary the lua command only works in single-player though? 03:46 Blockhead256[m] yes, it's disable for safety 03:46 Blockhead256[m] but I do more debugging in singleplayer 03:47 fluxionary does it have more security implications than just being an admin on a server? 03:47 fluxionary i guess if it's insecure environment and doesn't restrict that, ok, i understand 03:51 fluxionary i've also used the equivalent of print statements for debugging regularly, for lack of a stepping debugger. this certainly seems worth testing 03:58 muurkha a stepping debugger can be nice (I have one open right now!) but an awful lot of debugging is better done with printf 04:02 muurkha a stepping debugger is great for inspecting in detail the state of a program at a single moment of execution, or a few of them 04:03 muurkha but printf debugging shows you the entire history of an execution 04:03 muurkha I often find that the latter is more useful 04:06 muurkha perhaps because I more often know what variables or what kind of events are relevant to a bug than which moment in time is the one where it manifests 06:00 fluxionary_ re: but printf debugging shows you the entire history of an execution 06:00 fluxionary_ only if you manually instrument your own code. 06:02 fluxionary_ i've used tools for java, c, and python, which let you rewind and replay your code w/out much trouble 06:03 fluxionary_ i wouldn't be surprised if something similar existed for lua, but i haven't found it yet 06:07 fluxionary_ "print" isn't a great debugging mechanism, but sometimes it's the best you've got access to 06:09 fluxionary_ i've personally decided to add a "DEBUG" detector to my pre-commit script, so i can add debugging stuff to my code and reduce my chance of accidentally committing it 06:29 muurkha I keep planning to try out rr-project now that I have a CPU that supports it properly 06:29 muurkha my experiences with replay debugging in GDB were disappointing; the slowdown was something like a factor of 100,000 06:31 muurkha so you kind of need a nondeterministic bug to justify using that kind of replay debugging; it's almost invariably faster to just restart the program 100 times with breakpoints in different places instead of running it once with replay recording turned on and then reverse-executing to a series of watchpoints or whatever 06:32 muurkha what kind of tools for Java and Python do you mean? 06:32 muurkha you could use rr (or GDB) with a Lua program too 06:32 muurkha with respect to manually instrumenting code, I've often manually instrumented someone else's code for this kind of thing :) 06:33 muurkha but yeah breakpoint debugging definitely moves up the league table significantly when Other People's Code gets involved 12:48 MTDiscord mh is it just me or is the minetest repo missing a commit/PR to match the TextureLayer renaming commit in the irrlicht repo? 16:53 sfan5 yes 16:54 sfan5 use whatever version is documented in misc/irrlichtmt_tag.txt if you want a guaranteed working build