Time Nick Message 00:38 Zughy[m] meeting reminder in less than 24 hours 17:38 Krock 22 minutes until meeting 18:01 Krock It is time. Desour Zughy[m] sfan5 rubenwardy 18:02 rubenwardy here 18:03 Krock > Where should we post next meetings notice? (Zughy) 18:03 Krock (slowly ramping up) ideas? 18:04 Krock this is IMO a step back but I suppose we cannot do much else than to use the Discussions alternative 18:04 Desour maybe put all that belong to one major version into a separate article? 18:05 rubenwardy ? 18:05 Krock I think it's about the GitHub teams page that is about to shut down in May 18:05 Desour i.e. one wiki page for 5.7, one for 5.8 and so on 18:05 Desour ahh 18:06 Krock the changelog is fine as-is 18:06 Krock s/changelog/meetings wiki/ 18:07 Guest54 you could always make a repo with markdown/rst/asciidoc files and use that? 18:07 Krock teams pages were handy to core devs about upcoming meetings and to discuss security issues 18:07 rubenwardy https://github.com/minetest/discussions/discussions 18:08 Krock both are not suited for GitHub Discussions 18:09 Krock well I don't know. perhaps #minetest-staff will be again more used for such things? 18:09 rubenwardy Ok, I've moved it from team discussions to a private GitHub discussions repo 18:10 Krock hmm 18:10 Krock I see. 18:10 rubenwardy ah cool https://github.com/orgs/minetest/discussions 18:11 rubenwardy it's basically the same but stored in a different place 18:11 Krock it's a more convenient layout 18:16 Krock rubenwardy: do you see anywhere who has got read access to this page? 18:16 rubenwardy I added MTE + MTG + triagers 18:16 Krock alright 18:17 rubenwardy it's based on https://github.com/minetest/discussions/settings/access 18:17 rubenwardy https://rwdy.uk/2dgnr.png 18:17 Krock 404 on my side, as well as the page itself when trying to access without cookies 18:17 rubenwardy yeah guess it requires admin privs 18:17 Krock 404 is a bad error for this but I guess that's for security 18:18 Guest54 well, otherwise you'd be able to infer if something was there 18:18 Krock that's the purpose of error 403. as long you cannot see or change anything it does not matter 18:18 Krock moving on 18:18 Krock > Complaints about connected glass changes - changes the meaning of glasslike_framed_optional and there's no way for texture packs to control it. 18:19 Krock was there any consensus about this yet? 18:19 Guest54 it's not only that, it doesn't work for anything not 16×16 18:19 Krock I've seen a few suggestions to let texture packs decide but for an upcoming release that's out of reach 18:20 Krock "does not work" is solely due to the nature of the frame thickness 18:20 Desour Guest54: mods should test nodes with that drawtype with the setting on and off, so 16x16 not working is irrelevant here 18:20 Desour non-16x16* 18:21 Guest54 Krock it is not necessarily due to the frame thickness, some glasslike textures have fuzzy frames that work perfectly with the non-frame rendering, but would need something like a 9-slice connected glass to work with the frame one. 18:21 Desour if someone uses a texture pack, they can also change the setting. 18:22 Krock how many people have complained about this so far? 2? 18:22 Guest54 Desour we can agree on “should” all we want, it doesn't change that the default settings used to work and now they don't and no remedy is given to server or game devs who explicitly used this for “if you have a texture pack that works with it, enable it” 18:25 Guest54 Krock maybe “how many people have complained from the ones who tested this more than whoever approved that change” would be a better metric 18:25 Desour we could just revert the change and open the issue again, so complainers will shut up 18:25 Guest54 also note that the people who wanted connected glass the most dislike the frames 18:25 Guest54 they want something more like optifine 18:26 Guest54 i.e. frameless (9-slice?) 18:27 Krock well tbh I am getting tired of this discussion. usually there would be a few core devs around to give inputs to approve/disapprove but today it's dry as ever 18:27 Krock I'd be in for reverting. Opinions, Desour, rubenwardy ? 18:28 Guest54 would it help if i provide all the previous issues across different projects regarding the thing? (where it was always rejected) 18:28 Guest54 like, to evaluate the arguments 18:28 rubenwardy I don't feel to strongly either way, just as long as a decision is made 18:28 Krock Guest54: feel free to put them into #8290 18:28 ShadowBot https://github.com/minetest/minetest/issues/8290 -- Enable Connected Glass By Default 18:30 Krock there's also some IRC discussion which I no longer can find 18:33 Guest54 i am not logged in to github right now (i need to dust off a machine for it, will do soon), but here is the mtg discussion about framed obsidian glass as default (rejected): https://github.com/minetest/minetest_game/issues/1547 18:34 Guest54 and here is the discussion for separating glasslikes into framed and not framed in mtg: https://github.com/minetest/minetest_game/issues/2434 and here is another one https://github.com/minetest/minetest_game/issues/2326 18:34 Guest54 in general, it seems it comes up every few years and is rejected (except this time) 18:35 Guest54 the arguments are also the same every time and everyone agreed using optionally connected glass in many mods by default was a mistake … 18:35 Guest54 rubenwardy krock would it help if i volunteer to make an additional glasslike drawtype that is more like optifine? 18:36 Krock I like the current frames, though. 18:36 Krock not that it would prevent you from adding another mode, ofc. 18:36 Guest54 i would just add it so that people who want the connected glass can have it like optifine. i *think* PrairieWind is one of them, am i right? 18:37 Guest54 also Zughy[m] do you have an opinion as a texture creator? soothing32 sure does look different connected (i.e. largely invisble except for the frames) 18:38 Desour the current connected glass looks good. a separate type should not replace the current one. such a new drawtype would probably not be high priority => you can implement it of course, but don't expect it to be merged fast 18:39 Guest54 Desour i was more asking “if i pledge to make a connected glass drawtype to end the discussion once and for all in the future, would that convince anyone today to change the default back?” 18:39 Desour (with fast I mean minetest-speed fast, so weeks or months) 18:39 Guest54 i have been offline for months, i can wait 18:42 Zughy[m] Non connected glass is just ugly imho 18:47 Guest54 it's a question of taste, but generally it depends on your texture pack really. the handpainted style for example does not work connected as I see it, regardless of frame length. anyways, i doubt anyone is interested in the details of glasslike rendering much rn, because otherwise someone else would have complained much more than me. 18:47 Guest54 is there any other topic? 18:50 pgimeno what are development versions and feature freezes for, if not to revert changes that turn out to cause trouble, so they can be examined better in a latter iteration? 18:51 Krock based on the discussion from https://irc.minetest.net/minetest-dev/2023-03-26#i_6067064 I now revert #13242 18:51 MTDiscord tbh, it seems just one person is very vocally whining, is there anyone else who has raised concerns? 18:51 ShadowBot https://github.com/minetest/minetest/issues/13242 -- Enable connected glass by default by PrairieAstronomer 18:52 Guest54 that confuses me also, i thought reverting/disabling something that causes several problems at once before a release was business as usual (see shadows, debug info) 18:52 Krock @Jonathon That's also what I think but I'm getting tired of this 18:53 Guest54 Jonathon, you can try all matter of texture packs with it yourself. the argument is not “(non-)connected glass as more beautiful”, it is “the defaults used to work with everything and now they don't”. 18:56 Guest54 besides, contentdb should probably lint texture packs so that they do provide a usable special tile for connected glass, because if they are not, it's even uglier (try 7px with connected glass, it's arguably the fault of the texture pack author though) 18:57 Krock > 5.7.0 release 18:57 rubenwardy looks like it's just particle spawner stuff remaining 18:57 Krock right 18:58 Krock if it is wanted, I can reduce the modified code lines to keep the patch shorter 18:58 Guest54 in case anyone wants to implement ”proper” connected glass (i.e. frameless, with good creative texture support), 2 days ago someone pointed out in an issue that you need like 46 textures in minecraft to give the texture artist full creative control, so i was wrong about the 9-slice it seems 18:58 Guest54 https://github.com/minetest/minetest/issues/11546#issuecomment-1491504600 18:58 Krock but compatibility is generally easily tested by using clients only a few commits apart 18:59 Krock so if you have some inputs on these PRs it would be great to share them 18:59 Guest54 Krock can i help compat testing somehow? if so, where are the instructions? 19:00 Guest54 regarding release, i may have found a crash on a corrupted texture file, but i can't reproduce right now (something something me writing the wrong bits by accident, i was also able to crash sxiv doing that). would this hold up your release? if not, i'll not hurry about reproducing it. 19:00 Krock Guest54: instructions are here #13199 19:00 ShadowBot https://github.com/minetest/minetest/issues/13199 -- Particle spawner tween properties for old fields are not sent to clients 19:00 Guest54 thx 19:00 Krock the [png issue is already fixed FWIW 19:00 Guest54 no, another crash 19:01 Guest54 simply a malformed file because i messed something up, but i seem to have gotten rid of that file using ”git clean” or so by accident 19:02 Guest54 i.e. most i can say right now is “if you fuzz all texture loaders, you'll get a crash”, which isn't much (and also hard to do on short notice since the unit tests for irrlichtmt are gone) 19:02 Guest54 i'll report back when i have it 19:03 Krock great, thanks. 19:04 Krock is there anything else to discuss? 19:04 Krock urgent PRs, issues? 19:09 Desour should we do something about https://github.com/orgs/minetest/discussions/57 before the release? 19:09 Krock sfan's idea is good. I can submit a patch for that 19:10 Krock should be rather trivial 19:10 Guest54 maybe it's the wrong time to ask, but why do yo uhave private development discussions anyways? is this a sensitive topic, security or so? 19:10 Desour Guest54: yes 19:10 Krock this one in particular: yes. 19:11 Guest54 i see thx 19:22 rubenwardy the MTG fix is https://github.com/minetest/minetest_game/pull/3022 19:23 rubenwardy Guest54: private discussions are used for security discussions before they're made public, and for meeting notices (Which are also posted here) 19:31 Guest54 rubenwardy why does that fix not just strip STX and ETX? (i assume those to be the problematic bytes) 19:31 luk3yx Guest54: I believe it was because there wasn't any reason for them to accept other control codes 19:32 rubenwardy Guest54: not sure why you're asking me rather than @luatic 19:32 Guest54 well as long as it's not on an engine level … \0 i understand though 19:32 Guest54 rubenwardy oh sory 19:32 MTDiscord @Lars 19:32 luk3yx I did ask about them blocking DEL 19:32 luk3yx I'm blocking it because it's an ASCII control character and there's no reason to use those 19:33 luk3yx Why would you want DEL in a formspec field? 19:35 Guest54 ig as long as utf-8 mostly passes through, i'd not be worried about breakage 19:36 appguru yeah, I'm preferring to be overly cautious here 19:36 Guest54 i mean this only affects mtg, so i can still have a “unicode book” with all the control codes, right? no harm done. 19:37 appguru yeah, this is a MTG PR 19:37 appguru C0 control codes have no place in "text" 19:37 appguru (that said, of the C0 control codes, I only know that \0 and \27 can cause Minetest to behave funny; the former for obvious reasons, the latter because of color & translation escape codes) 19:37 Krock uuuh I cannot reproduce this issue. pasting more than approx. 200 KB on my system does not work 19:38 appguru It's not (just) about pasting 19:38 appguru Modified clients and netcat can be used to send malicious input just as well 19:38 appguru You'll probably have to modify your client (e.g. to take the first char of your msg and repeat it 8 million times) to reproduce this 19:39 Guest54 modified clients definitely already do send formspecs to exploit stuff. in mcl2 the formspec for ender chests was only sent to the client when the chest was opened – some funny people just submitted the formspec results without ever having seen the form, using a CSM. 19:43 Guest54 Krock here is a template for sending whatever you want that fleckenstein once wrote to do what i just described, you might be able to pre-fill it or so? https://github.com/dragonfireclient/enderchest/blob/master/init.lua 19:45 Krock #13380 19:45 ShadowBot https://github.com/minetest/minetest/issues/13380 -- Limit formspec fields to 640K by SmallJoker 19:46 Guest54 Krock does this matter for inventory formspecs of shulkers full of shulkers? 19:46 Krock item stacks are not formspec fields 19:47 Guest54 so this is only “no one should copypaste the entire history of mankind into a book” ig? 19:48 Krock well.. my system did not allow to paste a bit more than 200 KiB anyway 19:48 Guest54 rubenwardy would you maybe give me a download of cdb so i can try some semgrep things? e.g. looking for wrong usage of string:find() with last parameter not plain (that can go crashy fast) or finding dupes. i'll report what i find of course and how i fonud it. 19:50 Guest54 Krock i'd ask cora or some other admin of a server what's the longest legit formspec input they had instead of just eyeballing it. i mean … a sane limit might just be a lot slower. in the other block game i believe you can put 319200 bytes in a book (79800 characters × 4 bytes because lol astral plane), but i obv did not test that. 19:51 Guest54 lower, not slower 19:51 rubenwardy feels like it would be best to do that using the ContentDB API 19:51 Guest54 rubenwardy obviously, i just want to experiment to see if it is worth it 19:51 rubenwardy download all packages I mean 19:52 Guest54 tbh i didn't think it would be appropriate to just download HUGE amounts of data without asking you 20:05 Desour Krock: chat is already size-limited to 64ki due to packet serialization, right? 20:05 Krock depends in which direction you look at it 20:06 Krock client --> server should already be limited 20:06 Krock well "should". I cannot tell for sure 20:07 luk3yx There's a chat_message_max_size setting that limits client to server chat messages 20:08 Krock yet another somewhat useless setting that could be replaced by a hard limit f 500 characters 20:08 Guest54 Krock do unit tests for the network packet handling exist? 20:09 Krock yes 20:11 sfan5 @luatic can you fix this in MTG https://i.imgur.com/koDjSyg.gif 20:16 Krock oh my god. entirely unplayable 20:17 Desour yeah, "1/0" is past-the-end 21:13 Zughy[m] why the fuck "connected glass" has been reverted? 21:45 kilbith because of the prick called erlehmann 21:47 rubenwardy which window has glass behind it? https://rwdy.uk/vpmlb.png 21:50 MTDiscord is it possible to determine? 21:50 rubenwardy it's not 21:52 Pexin as intended. which demonstrates it should be determined by the server. as such, changing the default optional mode is a bit pointless at least. 21:56 MTDiscord I think that glass nodes should have some sort of "impurities" in that case 21:56 MTDiscord like glass scratches 21:58 Pexin normal glass already does 21:59 MTDiscord but it also has the ugly outline 22:03 MTDiscord Seems like glasslike_framed_optional was a mistake. I ended up deprecating it myself in my own game.