Time Nick Message 01:42 [MatrxMT] model-swapped carts :D 01:42 [MatrxMT] honestly not a bad idea if you don't have the CPU budget for real advtrains I guess... 02:06 cheapie Blockhead256: It's more fun swapping them into the motorbike mod, because then you can have a train driving down the road and occasionally pulling a wheelie 02:07 cheapie ...and I also have a script that turns them into functional lawnmowers 02:09 cheapie Like this, except imagine /this/ model-swapped: https://cheapiesystems.com/media/2025-04-26%2014-44-16.webm 02:10 [MatrxMT] yeah I've seen the lawnmowers, pretty funny 02:12 cheapie How about the musical barcode scanner? https://cheapiesystems.com/media/2025-04-16%2018-15-33.webm 02:13 cheapie Inspired by https://www.youtube.com/watch?v=aTiHkr1z6qQ 04:11 ltuyen When i was an kid, i remember watching CN to watch some shows, including Ben 10 :) 04:11 ltuyen The only good thing about CN is just nostalgic old shows, nothing else 04:13 ltuyen Now, I just don't have an Cable TV in my house. I don't need that. 06:16 repetitivestrain Why was https://github.com/luanti-org/luanti/issues/5916 closed? 06:16 repetitivestrain this is the fix: https://codeberg.org/halon/Minetest/commit/2dd6fab54ab06edef2663ad7e594e5287d3c0fb4 06:17 [MatrxMT] looks like it was closed because of the report of it not reoccurring from "IrrLichtMt4", when triager Zughy then closed some time later 06:18 [MatrxMT] if it's still actually occurring then that's important info 06:18 [MatrxMT] if I try to recall, I think I've seen that message when the server is under extreme load 06:20 repetitivestrain the really important ramification is that a MapChunk where just one block fails to generate will be reported in whole to on_complete 06:20 repetitivestrain on_generated callbacks* 06:20 repetitivestrain and when subsequently emerged again will prompt the entire MapChunk to be regenerated 06:21 repetitivestrain Blockhead256: also, these were IrrLichtMt4's exact words: _the issue doe not happen as much_ in the recent version of Minetest but it _still happens_. [...]] 06:22 repetitivestrain it is produced by MapBlocks queued for emerge being unloaded before the emerge process completes 06:22 [MatrxMT] true, I was only glancing, but basically my point is I figure the same level of fuzziness I just displayed was probably at work when they closed it.. 06:22 repetitivestrain which is effectively addressed by acquiring a reference to such MapBlock for as long as they are in generation 06:23 repetitivestrain i am only asking because this is critical to the functioning of mineclonia's new map generator on slower PCs 06:23 [MatrxMT] would you like me to send that patch upstream? 06:24 repetitivestrain That'd be lovely 06:24 repetitivestrain thanks 06:25 [MatrxMT] can you please fill out https://raw.githubusercontent.com/luanti-org/luanti/refs/heads/master/.github/PULL_REQUEST_TEMPLATE.md and send me a pastebin? 06:25 repetitivestrain Sure, 10 minutes 06:34 repetitivestrain Blockhead256: https://paste.debian.net/1385796/ 06:52 [MatrxMT] hopefully this is right https://github.com/luanti-org/luanti/pull/16335 06:56 repetitivestrain appears splendid, thank you 06:57 [MatrxMT] not sure about GitHub ToS, but wouldn't it be nice to have a bot that takes an HTML form and submits it as a PR? anyway, would have to be hidden or something, would probably get spammed 06:57 repetitivestrain Is Wuzzy here? 06:57 repetitivestrain Would he test https://codeberg.org/halon/Minetest/commit/d2106ed002e4a833e35bdb51fc557b83f906feba and report whether it resolves https://github.com/luanti-org/luanti/issues/16197? 06:58 repetitivestrain i wrote this commit specifically for my CSM (where the entirety of the LocalPlayer physics step is implemented in Lua) but it's not immediately apparent that only my CSM is impacted by this tolerance issue 07:00 Blockhead256 not in the client list, so you'd be counting on him checking the logs 07:00 repetitivestrain Regrettable 07:00 repetitivestrain why hasn't he a bouncer... 07:00 [MatrxMT] you could always try a !tell 07:01 [MatrxMT] https://wuzzy.codeberg.page/meta/contact/ 07:01 [MatrxMT] may or may not be of use 07:01 repetitivestrain the email should be 07:12 repetitivestrain i can reproduce the abrupt false horizontal collisions in devtest too 07:13 repetitivestrain by simply digging a two block tall tunnel, rapidly jumping while moving forward, and observing that the X and Z velocities are sporadically zeroed despite the absence of any horizontal collisions 08:53 [MatrxMT] https://forum.luanti.org/viewtopic.php?t=31828 - why *don't* we support unicode filenames?? 10:07 sfan5 portability and simplicity. and this is unlikely to change. 10:09 [MatrxMT] we have unicode support in the locale file formats, is it because there might be problems like UTF16 filenames on certain OSes? 10:09 sfan5 handling UTF-8 needs extra thought when working with file transfer, url encodings, network encoding, ZIP files and more. while keeping a safe set of ASCII is just guaranteed to work. 10:10 sfan5 naming your media file 水.png is all cool until someone uploads a ZIP file with https://en.wikipedia.org/wiki/GBK_(character_encoding) to contentdb and it stops working for no apparent reason 10:10 sfan5 or https://en.wikipedia.org/wiki/Shift_JIS or any of the stupid latin variants that we had before unicode 10:11 sfan5 realistically Luanti's codebase is well prepared to handle this and it might even work as-is (when removing the restrictions) but there a large potential of bugs that make it just not worth it 10:12 [MatrxMT] it would also introduce a network incompatibility again right? 10:13 sfan5 sort of. any server using unicode media filenames would not work on older clients 10:15 repetitivestrain sfan5: ... if an emerge is aborted this needs to happen too, but I don't see any code for that. 10:15 repetitivestrain where are such aborts initiated? AFAICT getBlockOrStartGen calls initBlockMake, and excepting Lua crashes (in which case the server aborts anyway), the emerge thread will proceed to completion and invoke finishGen 10:16 repetitivestrain which invokes finishBlockMake to release the acquired reference 10:17 repetitivestrain if by ``abort'' you mean EMERGE_CANCELED, then isn't it impossible for this condition to be reported if initBlockMake succeeds 10:19 sfan5 that is true but I'd rather have the code handle the lua error case cleanly to avoid the risk of broken refcounting if someone edits the code in the future 10:20 repetitivestrain alright, i'll adjust it when i get off work 13:26 repetitivestrain sfan5: how's this? https://codeberg.org/halon/Minetest/commit/e93848da06a5d10c74d21a6a82983a0699eb588e 14:07 MTDiscord repetitivestrain: btw, thanks for the info yesterday, that's interesting. i wonder, if luanti were on codeberg, would you directly interact with the issue tracker? 14:07 repetitivestrain luantic: i would, yes 14:07 repetitivestrain Codeberg satisfies the FSF's ethical repository criteria and it doesn't require me to surrender my number for 2FA 14:08 sfan5 neither does github btw 14:08 repetitivestrain luatic: also i apologize for having misspelt your handle 14:09 sfan5 repetitivestrain: aside from the code style, sure 14:10 repetitivestrain sfan5: i'll correct that also in a trice 17:15 krikoepaf How to build Luanti without SDL? 17:16 Krock krisbug: -DUSE_SDL2=0 17:16 Krock oops. wrong autocorrect 21:05 ireallyhateirc @luatic, why are you reviewing our PRs in Exile? :P 21:05 ireallyhateirc I mean I don't mind it, we just weren't expecting you there 23:38 MTDiscord i was just looking at things in the luanti discord, saw a link to that PR posted there, peeked at it, and noticed some things :p 23:58 ireallyhateirc I see