Time Nick Message 03:27 [MatrxMT] are the distros keeping people back? are we just not selling people on new versions well? 03:29 MTDiscord Not to bring my own prejudice into it, but minetest might just be nicer to type :itnaul: 03:30 MTDiscord No the all change happened after 5 9... I'm sure it's probably a safe bet to blame Canonical/Ubuntu, probably just do that. 03:32 MTDiscord Couldn't be Debian's fault, they're notorious for brushing packages through without testing... or no wait that's Arch?. 03:32 MTDiscord *rushing 03:45 [MatrxMT] debian stable is on 5.6 though, whereas there are players on 5.7-5.11 03:45 [MatrxMT] trixie probably isn't getting 5.12 though, I think it's in freeze now (I think) 03:46 [MatrxMT] I type `lt` to launch the game, $ ls -l $(which lt) 03:46 [MatrxMT] `lrwxrwxrwx 1 root root 21 2024-11-15 16:41 /usr/local/bin/lt -> /usr/local/bin/luanti` 03:46 MTDiscord You also have to remember death of the release schedule switched from bi-yearly to quarterly, so just off of that it's going to look worse then it actually is. 03:46 MTDiscord Debian is so stable that pentium 4 is the newest supported cpu 03:46 [MatrxMT] boo, get new material 03:47 MTDiscord I would but my material is also stable 03:47 MTDiscord Translate("it works on my machine") => "It executes on my SmartWatch" 03:48 MTDiscord The joke is that Android uses the Linux kernel. 03:48 MTDiscord And I think BusyBox too 03:49 MTDiscord Or it might be ToyBox now, it's been a while since I had a root shell on a phone. 07:38 repetitivestrain sfan5: appears to work, thanks 07:40 repetitivestrain Btw, there is another memory leak capable of occupying 20+ GiBs in mere seconds when voxelmanips are transferred as parameters to async calls after automatic scaling has created more than one async environment 07:40 repetitivestrain i'd be glad to share a reproducer 07:44 repetitivestrain what is more, has anyone considered this proposition of mine i posted to the forums last week? https://forum.luanti.org/viewtopic.php?t=31670 07:46 repetitivestrain since singlenode and therefore all lua mapgens consider any unavailable block above sea level to be a source of sunlight for propagation (and contrariwise if below sea level), and it is impossible to alter this conception of sea level, caverns that generate across y=0 tend to be quite bizarrely illuminated 07:47 repetitivestrain and enabling mapgens optionally to generate a complete vertical section of the level at once would obviate the need to make assumptions as regards lighting in unloaded blocks entirely 08:41 repetitivestrain sfan5: https://www.reddit.com/r/Minetest/comments/1kvz8jq/working_on_a_mod_to_make_modular_entities/ ... 08:42 repetitivestrain well you might be interested in a similar twisting and modular eldrich abomination implemented with modern 5.9.x+ bone overrides 08:42 repetitivestrain https://codeberg.org/mineclonia/mineclonia/src/branch/main/mods/ENTITIES/mobs_mc/ender_dragon.lua 08:44 repetitivestrain twisting, vermicular, and modular, to be precise 11:44 Juri One user just updated to 5.12 and complained that his graphics look bad now 11:45 Juri Turns out he used bilinear/trilinear filtering and liked the blurry look 11:45 Juri He was using texture_min_size=16 and complained that the min value now is 192 11:46 Juri I don't fully understand how the settings work together, but my question is, is it possible to get the old blurry style back with settings? 11:58 sfan5 Juri: no, it's 100% intentional that stuff isn't blurry anymore 11:59 Juri Yep I figured from reading PR 16034 11:59 Juri Thank you :) 12:00 sfan5 repetitivestrain: if you want to get a guaranteed response to something open an issue on the luanti repo 12:04 MinetestBot 02[git] 04sfan5 -> 03luanti-org/luanti: Fix texture double-free in main menu 13da7897a https://github.com/luanti-org/luanti/commit/da7897a822e37a7b373e49f135318369c5c3ba17 (152025-05-27T11:56:18Z) 12:09 repetitivestrain sfan5: i don't want to register a github account for myriad reasons 12:10 repetitivestrain if anyone's amenable to submitting an issue on my behalf that would be super and i'll send you a reproducer 12:29 Swissa1pS "I'll send you a reproducer" maybe start with that. post it along with the usual description for the issue, then drop the link here and/or the forum. Otherwise it is very unlikely that anybody will pick it up for a myriad of reasons ;P 12:31 Swissa1pS asking others to do what you are unwilling to do - rarely goes anywhere 12:46 [MatrxMT] i could do it 13:02 sfan5 repetitivestrain: well then forum is better than not at all. alternatively see if you find someone who can post it for you 13:02 sfan5 as for the memory leak just throw me the example code 13:45 repetitivestrain sfan5: https://paste.debian.net/1376901 13:46 repetitivestrain with one async environment rss plateaus long before 10 GiB, but with more than one it expands boundlessly or at least till my machine begins thrashing 13:50 repetitivestrain but https://codeberg.org/mineclonia/mineclonia/pulls/3419 is really the best reproducer--just modify AsyncEngine::initialize to initialize four or so workers at startup and observe the postprocessing stage of this mapgen rapidly consume 20 GiB+ of rss 13:51 repetitivestrain while it remains alternates between 2.5 GiB and 6.0 GiB without 13:57 sfan5 that's easy to explain 13:57 sfan5 it's just https://github.com/luanti-org/luanti/issues/13982 13:58 sfan5 to fix this you have to explicitly free the voxelmanip in both environments 13:59 sfan5 I guess there's a problem if you want to return a vmanip from an async callback. you don't have an opportunity to free it. 14:00 sfan5 you could work around this, sort of 14:03 repetitivestrain sfan5: that is what i assumed initially, but i couldn't account for heap consumption being so much worse with two active async environments than with just one 14:03 sfan5 🤷 14:04 repetitivestrain hence there must be something more sinister behind so enormous a difference as the difference between a few more GiB than would be ideal and thrashing 14:06 sfan5 I wouldn't investigate this more before you have confirmed that doing vmanip:close() doesn't fix it 14:06 [MatrxMT] wait is async mul5ithreading or what 14:06 repetitivestrain sfan5: i test with my own patch (which is `vm:clear ()' rather than `vm:close ()') 14:06 repetitivestrain tested, i mean 14:07 repetitivestrain but the two functions are substantively identical 14:07 repetitivestrain https://codeberg.org/mineclonia/mineclonia/src/commit/mineclonia_mapgen/mods/MAPGEN/mcl_levelgen/post_processing.lua#L804 14:07 repetitivestrain the true arg to vm:write_to_map calls `vm->clear' after the data is written 14:08 sfan5 I am talking about your minimal example 14:08 sfan5 there are too many unknowns to reason about the other code 14:14 repetitivestrain sfan5: that doesn't help 14:14 repetitivestrain clearing the VM, that is 14:15 repetitivestrain both in the async callback and after it is copied by handle_async 14:15 repetitivestrain rss climbed to 10 GiB, whereafter i had to terminate my experiment as my computer came perilously close to trashing 14:16 repetitivestrain thrashing* 14:55 sfan5 repetitivestrain: I didn't look closely and your example is actually the "there's a problem" case I described 15:10 sfan5 repetitivestrain: here's the (theoretically) leak free version https://paste.debian.net/hidden/c8222e79/ 16:49 MinetestBot 02[git] 04SmallJoker -> 03luanti-org/luanti: Formspec: Fix incorrect cell size when using non-default fonts (#16178) 1394a9b94 https://github.com/luanti-org/luanti/commit/94a9b94baf5c6626b1f917ef2007e760786401e8 (152025-05-27T16:47:41Z) 17:17 [MatrxMT] I just noticed on arch's pacman the info about luanti says it is infinite-world. Wasn't there a limit? 17:18 [MatrxMT] yes, -31k to +31k, not sure why it says infinite... 17:19 jonadab Someone probably made an assumption when writing up the blurb. 17:20 jonadab In practice, the only world edge anyone ever runs into (except deliberately as a stunt), is the bottom. 17:20 jonadab So anyone who doesn't actually know, might guess that it's infinite. 17:20 [MatrxMT] How far would the player even dig?! 17:21 jonadab Depends what modules you have installed. 17:21 [MatrxMT] valid 17:21 jonadab If you've got things like caverealms and dfcaverns at various depths, it can be worthwhile to dig down most of the way to void. 17:22 [MatrxMT] How does IRC work? 17:22 [MatrxMT] Did I understand it correctly you need to run your pc 24/7? 17:22 jonadab (And then you install travelnet elevators or something.) 17:23 jonadab Technically, you don't _need_ IRC to run 24/7, but most of us do leave it running in a screen session, so we can just connect to the session whenever we want; it's convenient. 17:24 jonadab Some people use a "bouncer" (IRC terminology for a proxy), and then connect to that. The bouncer often runs on a headless system or VPS. 17:24 jonadab People who use IRC _tend_ to be IT professionals or otherwise technically inclined, and _tend_ to have computers that are on 24/7 anyway for various reasons. 17:29 [MatrxMT] why use IRC anyways? What are the benefits? 17:29 MTDiscord the reason for using an IRC bouncer that runs 24/7 is usually to keep receiving channel history while you are not connected. you connect to the bouncer and you'll receive all the messages you missed 17:30 MTDiscord ...but nowadays a lot of IRC channels you may want to be in do public logging or bridge channels to discord or matrix which have their own inherent value as a pseudo-bouncer 17:34 MTDiscord unfortunately IRC is kinda dead nowadays and mostly filled with bouncer zombies. :( used to be active on IRC for some years before most got absorbed into the discord singularity around 2017-2018, but even then it was still niche 17:36 jonadab Not sure how to interpret the question, about the benefits of IRC. As compared to what, passenger pigeons? 17:36 jonadab Mostly I use IRC because it is the dominant real-time chat protocol for most of the topics that I'm interested in discussing. 17:37 jonadab And most of the alternatives are proprietary protocols with no decent configurable client. Discord, for example, is horrific, I wouldn't use that if they paid me. 18:01 [MatrxMT] irc.luanti.org 20:36 [MatrxMT] woah what a format