Time Nick Message 11:02 Mowgli Hi.. 11:02 Lesha_Vel hello 11:03 Mowgli I am here because of bug #13381 11:04 Mowgli The main problem is that with on_use defined, the repeating action is disabled. 11:04 Mowgli How to reenable it again? 11:06 Mowgli Anybody here who have an idea? :-) 13:01 independent56 It's nice to see more and more innovation in Minetest over time 13:01 independent56 Technic and Pipeworks were innovative at their time and after a few years, new mods pushed further. Techage, Advtrains, Ladybug game thing, and others are all pushing the boundaries of what's possible 13:02 independent56 With the rise of AI and the community, it might grow exponentially 13:04 independent56 I'm currently teaching myself techage and it feels like Pipworks, Technic, Meseconds, and various Mesecons_addons are all being mixed together in a cohesive system. 13:06 independent56 There are even parrelels to digilines 13:09 independent56 Ugh, i feel like my mesecons, Technic, and digilines knowladge has all been obsoleted 13:09 independent56 First, ChatGPT and now this? 13:21 independent56 !tell Yad, Fix your connection please 13:21 MinetestBot independent56: I'll pass that on when Yad, is around 13:21 independent56 !tell Yad , Fix your connection please 13:21 MinetestBot independent56: yeah, sure, whatever 13:21 independent56 whoops, trailing comma 13:22 independent56 !tell Yad_ , Fix your connection please 13:22 MinetestBot independent56: I'll pass that on when Yad_ is around 13:22 independent56 Wrong username! What tough luck i must have! 13:26 MinetestBot Yad: Apr-08 13:21 UTC , Fix your connection please 13:29 MinetestBot Yad_: Apr-08 13:22 UTC , Fix your connection please 13:29 independent56 Haha 13:29 independent56 sorry lol 13:29 independent56 I only rejoined once; you about erm... 15 times. 13:35 Guest54 !tell Yad +++ATH0 13:35 MinetestBot Guest54: I'll pass that on when Yad is around 13:36 Guest54 i found the holy grail in hyperrogue! 13:36 Guest54 turns out that the sequence hex-hex-hept-hex-hex-hept is almost a straight line 13:36 Guest54 and if you construct a perpendicular to a chord in a circle on the hyperbolic plane, you can iterate yourself towards the grail 13:37 Guest54 https://old.reddit.com/r/Hyperrogue/comments/ooliaw/just_tried_to_find_the_holy_grail_for_like_45/ 13:37 Guest54 well, it's an equidistant curve of a straight line 13:39 Guest54 since it took me a long time to find the grail (it is in the middle of a circle with radius 28, but the circle is on the hyperbolic plane), I am now 110% convinced that users that ask for non-euclidean geometry are either pretty masochistic players of games … or have no idea what non-euclidean means 13:40 Guest54 Kimapr you should try hyperrogue, it is less difficult than nodecore skyhell, but every mistake is entirely your own fault 13:41 MinetestBot Yad: Apr-08 13:35 UTC +++ATH0 13:41 Kimapr i did play hyperrogue 13:41 Guest54 how far did you get? 13:41 Guest54 i rescued the princess today and found the holy grail 13:42 Kimapr hmm i didn't do any of those things 13:42 Kimapr idk 13:42 Guest54 did it appeal to you, being a big mean riddle? 13:42 Guest54 also do you have tips for other extremely masochistic game experiences? 13:42 Guest54 dwarf fortress does not count, because there, again, you are not responsible for every mistake 13:43 Guest54 i believe battle for wesnoth kinda counts. after all, people whining about “unfair RNG” are actually saying they can not handle the odds. 13:44 Guest54 Kimapr did you beat the plane of eternal running where the floor falls down? if you collect 10 phoenix feathers there you get the orb of safety, which allows you to save your game and escape enemies in the same instant 13:44 Guest54 it is one of the starting planes 13:45 Kimapr hmm i think i remember something like that 13:45 Guest54 the interesting thing is that because of the hyperbolic geometry you should always be able to escape the dogs 13:45 Guest54 they literally can not reach you 13:45 Guest54 because their path gets longer as the floor falls down behind you 13:47 Guest54 i wish minetest had strange floor tilings hehe 13:48 Guest54 wang tiles in minetest when 13:50 Goobax hello 13:50 Guest54 hi 13:51 Goobax who wants to help me to develop my minetest server 13:52 Guest54 what do you need help with? 13:52 Guest54 if it is a small task, i may be able to give advice, but i can not guarantee it 13:53 Goobax I'm hiring developer 13:54 Guest54 Goobax why do you need to hire someone? is the task very complex? 13:54 Guest54 or a lot of work? 13:54 Goobax yes 13:54 Goobax a lot of work 13:54 Guest54 maybe you should describe what you need first. i thought you want to develop mods or so. 13:55 Guest54 in case you need an independent implementation of some minetest stuff, i suggest to look into the work that anon5, himbeerserver_de and fleckenstein did 13:56 Goobax I would like to create my own inventory not the inventory of minetest on this inventory I would like to put a faction menu with the homes and the recipes 13:56 Guest54 you mean you need help making a formspec? 13:56 Goobax it's what 13:57 Goobax ? 13:57 Guest54 formspec language allows you to specify a kind of dialog window that the client sees 13:57 Guest54 for example the inventory dialog 13:57 Guest54 or help dialogs 13:57 Guest54 or furnace interface 13:57 Guest54 you need that? 13:57 Guest54 have you checked out existing inventory mods? 13:59 Goobax yes I'm watching I don't like it 14:02 Guest54 btw, cora and me yesterday worked out last night why convert(1) from imagemagick flips images upside down (e.g. any TGA image with a scanline order bottom-top is flipped) and what to do about it. you have to pass -auto-orient to imagemagick. 14:02 Guest54 this actually triggers the bug in imagemagick twice: the image is mistakenly flipped, then mistakenly flipped again 14:02 Guest54 which means it is right side up 14:04 Guest54 the reason for the bug is that imagemagick developers map both scanline order and orientation data (e.g. exif) to the same data structure. so if you have an image format in which the scanline order is not fixed (AFAIK both TGA and BMP have this option), half of possible images are mistakenly flipped upside down on conversion. 14:05 Guest54 this is, of course, extremely stupid, but bugs against imagemagick were so far answered with “the previous behaviour was a bug and we fixed it“ 14:05 Guest54 that every other image processing tool outside of imagemagick is able to distinguish orientation information and scanline order must look like a bug to them as well ;) 14:06 Guest54 -auto-orient seems not to work for montage(1) though. i think it is because at that point the images are mistakenly flipped thrice, once during blitting. 14:06 Guest54 but that is only a suspicion, i have not dived too deep into it 14:06 Guest54 anyways, if convert(1) flips your image upside down, use -auto-orient to trigger the flip twice 14:07 Guest54 that's still a bug to everyone except the imagemagick developers 14:07 Guest54 but it mostly works lol 14:08 Guest54 in fact, because earlier, non-buggy imagemagick versions would not do the mistaken mapping, they should not flip anything at all with -auto-orient – which looks the same as flipping upside down twice obv 14:08 Guest54 i also investigated texture atlases 14:09 Guest54 and it turns out that using a texture atlas is a much better idea than putting 2000 PNG files in a zip 14:10 Guest54 because each PNG file in the zip has about 64 byte overhead minimum + the PNG data format overhead 14:11 Guest54 the difference between texture atlases and separate files is so huge that zipping all letters for mcl_signs as optimized PNG makes a 48kb zip, zipping them as TGA makes a 38kb zip … but uncompressed TGA texture atlas size is 14kb and PNG texture atlas size is 1.4kb 14:12 Goobax can you help me? 14:12 Guest54 this basically means that almost everything i have read about texture optimization for minetest in terms of both transfer and storage size is wrong by about an order of magnitude 14:17 Guest54 oh, also a zlibbed tga texture atlas is about the same size as an optimized png texture atlas, but is much faster to create 14:21 Guest54 i also found that the PNG footer is constant and thus its checksum is also a constant. hilarious. 14:29 independent56_ Why are all the modders so much more capable then me? What should i do to outcompete them? 14:29 Guest54 independent56_ the only winning move is not to play 14:29 independent56_ https://content.minetest.net/users/joe7575/ It's absolutely rediculous people can get this far whilst i am entirely unable to! 14:29 independent56_ Guest54, Yes but i have to play somewhere. 14:29 Guest54 independent56_ this is not a competition 14:29 Guest54 independent56_ if you want to make a cool mod, just do it 14:30 Guest54 independent56_ do you want mod ideas on what to implement? 14:30 independent56_ Yes but the problem is i don't have long-lasting motivation whatsoever 14:30 MTDiscord Every awesome mod someone else makes is an awesome mod you still get to use, but do not need to put in one second of coding time! 14:30 Guest54 hehehehe 14:30 independent56_ oh ok then 14:30 Guest54 independent56_ if you do not have long-lasting motivation, make smaller mods 14:31 independent56_ Yes but then i have to maintain them 14:31 Guest54 independent56_ one of my more popular mods is bushy leaves, which just changes leaves to render as a nice nodebox 14:31 independent56_ intresting 14:31 Guest54 independent56_ https://content.minetest.net/packages/erlehmann/bushy_leaves/ 14:31 independent56_ I'll look into such mods 14:31 Guest54 it is very small, but over 3000 downloads 14:32 Guest54 or this one, which changes the tip of a lever to red if the lever is on: https://content.minetest.net/packages/erlehmann/mcl_lever_status_indicator/ 14:32 independent56_ good idea 14:32 MTDiscord In this case you could contribute to someone else's mods. Then you still get to code, at your own pace, but the boring administration tasks is left to the maintainer. If you're searching for things to do, we have a bugtracker full of tasks on my server and most other maintainers are also happy if you hand in PRs for their issues. 14:32 Guest54 meanwhile, xmaps, which renders ingame maps, has only 700 something downloads https://content.minetest.net/packages/erlehmann/xmaps/ 14:33 independent56_ That's also intresing 14:33 Guest54 independent56_ do you need ideas for small mods? 14:33 independent56_ No, i can make them myself 14:34 Guest54 what ideas do you have? 14:34 independent56_ But then of course, why should i contribute to mods when i can contribute to the minetest engine? 14:34 independent56_ Navigating large mods is someitmes more difficult then an engine curated by a large community 14:35 Guest54 mods are easier to write 14:35 Guest54 IMO no mod should be particularly large ;) 14:35 independent56_ why not? 14:35 independent56_ Techage is perfectly fine being large? 14:36 Guest54 techage is a mod pack 14:36 Guest54 several mods together 14:36 Guest54 independent56_ if you want to do something that is quickly done: 3d ladders. i.e. ladders like they look in repixture, but for normal minetest game 14:36 Guest54 i bet people will use that 14:36 Guest54 i would, for example 14:36 Guest54 but i am not motivated enough to make it myself ;) 14:37 independent56_ Fine then 14:38 independent56_ It seems quite simple; steal it from repixture and rebrand it 14:38 independent56_ just like store-brand crisps 14:38 independent56_ (which take from brands) 14:38 appguru Guest54: This already exists, see https://content.minetest.net/packages/Linuxdirk/redef/ 14:39 Guest54 appguru nice thanks 14:39 Guest54 appguru look at the logs, i found that texture atlases are unreasonably efficient lol 14:39 appguru yeah I saw lol 14:39 independent56_ damn 14:39 independent56_ give me another idea 14:39 appguru Interestingly artists also seem to like working with atlases 14:40 Guest54 appguru not accoring to the mcl2 fanclub! according to that crowd, artists prefer working with a directory of 2300 small png files. 14:40 Guest54 (when zipped, each file has about 100 byte overhead, due to zip + png) 14:40 appguru It would probably be useful if there was a tool for artists to easily map regions of the image to (conceptual) textures 14:41 Guest54 texture editors are good enough as far as i know 14:41 appguru and then in the code you'd have a function T such that T"texname" would simply resolve to the appropriate [sheet texture modifier to extract the relevant portion from the atlas 14:41 Guest54 that's not that simple 14:41 Guest54 i mean you can do it in lua prob 14:42 Guest54 i also looked into actually supporting unicode using texmods, but i think rendering the text directly is a bit better 14:42 Guest54 it can work, by using 2 large texture atlases of about 2MB 14:42 Guest54 one for plane 0, one for plane 1 14:43 MinetestBot 02[git] 04RodentOfUnusualSize -> 03minetest/minetestmapper: Fix `dumpnodes` crash on deprecated tile field name 137af222d https://github.com/minetest/minetestmapper/commit/7af222dd9de079fc0efb8a7138997ea45da9d5e8 (152023-04-08T14:32:29Z) 14:44 Guest54 appguru could you maybe make a PR for xmaps to show me how you would handle the callbacks? 14:44 Guest54 i want to make xmaps an API with pluggable renderers 14:44 Guest54 and for that i need it to work lol 14:45 appguru sure 14:45 Guest54 thanks! 14:48 Guest54 appguru did you see my musings about filtering control characters? 14:48 independent56_ Guest54, throw another idea my way 14:49 Guest54 independent56_ make a nodebox that replaces each node with “sponge” in the name with a menger sponge fractal https://en.wikipedia.org/wiki/Menger_sponge 14:49 appguru Guest54: not sure, could you link them again? 14:49 Guest54 appguru i can repeat them here, hehe 14:49 independent56_ How deep? 14:49 Guest54 independent56_ whatever looks good 14:49 Guest54 independent56_ make a nodebox mess 14:50 Guest54 appguru basically, you need to filter everything that messes with lua at the input, after decoding the data, but everything that messes with the terminal at the output, after encoding the data. 14:50 appguru ah that 14:50 Guest54 appguru filtering ANSI escapes at the input is not only conceptually wrong, it lures you into a false sense of security. it also requires that you check every input for every output. 14:50 independent56_ Intresting idea. Should i keep the same texture? 14:51 appguru Guest54: Lua handles byte strings just fine. 14:51 Guest54 independent56_ no idea. look at the sponge in nodecore to see how it could look 14:51 independent56_ No, the same texture as the original node 14:51 Guest54 appguru well i thought you can do table injection using \02 and \03 or so? 14:51 appguru It's Minetest which has problems with null terminators, and possibly \27 (ANSI escape) depending on how the strings are used. 14:51 appguru Table injection? 14:52 Guest54 appguru which problem happens with escape? 14:52 appguru Guest54: Minetest uses escape sequences for translations and colors 14:52 Guest54 where? 14:52 appguru chat for example 14:52 appguru but also formspecs 14:53 appguru most (all?) places where text is shown 14:53 Guest54 ah, then i was mistaken about it only mattering in terminal output! 14:53 appguru Guest54: I think for terminal output ideally %q should be used 14:53 appguru But anyways 14:53 Guest54 appguru that syntax is gnarly 14:54 appguru I'm not proposing stripping escape sequences at an engine level 14:54 Guest54 nope 14:54 Guest54 they are needed after all 14:54 appguru But for processing inputs which should only be text, it is a sane thing to do 14:54 appguru Text may not contain control characters, period 14:54 appguru of course stripping control characters alone doesn't magically make the text "safe", but it at least makes it slightly more secure 14:55 Guest54 appguru oh also my coworker and me talked about control characters. turns out that you should probably not filter those that are not known to be malicious. and if you filter them, you need to act on the *decoded* codepoints, not the input bytes. i hope you do that? 14:55 appguru safe to pass around* 14:55 Guest54 i strongly advice to reject known-bad inputs and not sanitize them, see LANGSEC 14:55 Guest54 advise 14:55 appguru Guest54: Well, rejecting is more work :P 14:55 Guest54 usually it is not lol 14:55 appguru fair 14:55 Guest54 but in this case, your solution of stripping escapes makes a mess of the rest of the escape sequence 14:56 appguru yes 14:56 Guest54 you probably want to strip the entire escape sequence 14:56 Guest54 it's not too hard 14:56 appguru no 14:56 Guest54 they start with escape 14:56 Guest54 why not 14:56 appguru you aren't supposed to enter escape sequences in text input fields 14:57 Guest54 how do you know i am not? after all, the code allowed it ;) 14:57 Guest54 anyways, if escape sequences are what you want to protect against, you need to scan for escape sequences 14:58 Guest54 and btw, terminals usually ignore unknown escape sequences AFAIK 14:58 appguru as for filtering out only the malicious control characters, yes, I let them keep \t and \n 14:58 sfan5 I would say "PRs welcome" but a PR would not actually be welcome, nitpicking this is a waste of time 14:59 Guest54 well, incorrect filtering has introduced security vulnerabilities in the past, that's why i am saying it 14:59 Guest54 not in minetest as far as i know 14:59 appguru nobody is claiming that this filter is 100% secure 14:59 Guest54 but like, edge or internet explorer had an XSS filter … that transformed non-XSS payloads into XSS 14:59 appguru it's just better than... no filter 14:59 Guest54 yeah but i am claiming you can make it 100% secure regarding your attack scenario 15:00 appguru arguably I do not operate at a UTF-8 level; doing so would effectively require involving https://github.com/appgurueu/modlib/blob/master/utf8.lua 15:00 Guest54 does minetest only use C0 control characters or also C1? 15:01 appguru Guest54: I think just stripping \27 is reasonable. This will effectively disable Minetest and ANSI escape sequences, without having to add more logic to specifically remove those escape sequences. 15:01 Guest54 appguru they can be caught by a simple regex though if i am not mistaken 15:01 appguru Well too bad we don't have RegEx. 15:01 Guest54 like sed -e 's/\x1b\[[0-9;]*m//g' 15:01 appguru (Lua patterns aren't full RegEx) 15:01 Guest54 removes all colors 15:02 Guest54 unless i am talking out of my ass here 15:02 appguru I'm not sure whether Lua patterns would be sufficient 15:02 appguru But either way, I don't see a reason to overcomplicate this 15:02 Guest54 you can scan for escape [ %d+ m 15:02 appguru I might as well filter only \0 15:02 Guest54 btw, why is color in this case a problem? 15:03 appguru color isn't really a problem 15:03 Guest54 the sequence CSI %d+ m is used for color, right? where CSI is escape [ 15:04 Guest54 appguru what is it that you are filtering then? 15:04 appguru Minetest (translation & color) and ANSI escape sequences (not limited to color) 15:04 Guest54 oh what does translation use? 15:05 appguru \27(T@something)...\27E or something IIRC 15:06 Guest54 well, i think code should not only be correct, but obviously correct. i.e. it should state its purpose. but i see that it is more difficult to write a pattern (though i think that patterns are powerful enough) 15:06 appguru anyways, back to xmaps 15:06 appguru do you want me to kick players who can't receive dynamic media? ^^ 15:06 Guest54 anyways, just in case you want to do a more general filtering on engine level: please operate on the correct level (bytes vs codepoints) and filter at the correct stage (input vs output) 15:06 Guest54 appguru what the hell no 15:07 appguru hehe thought so 15:07 Guest54 why would you 15:07 appguru to make them rejoin 15:07 Guest54 so they receive the media? 15:07 appguru yes 15:07 appguru (and to bully them into upgrading) 15:07 Guest54 look i put a map on the wall and someone using minetest 5.3 comes along and gets kicked 15:07 Guest54 rejoins … and gets kicked again? 15:08 Guest54 or how do you envision it 15:08 appguru if the map updates, yes lol 15:08 Guest54 thanks, i hate it 15:08 appguru mission accomplished ^^ 15:08 Guest54 wait, minetest 5.3 has dynamic media 15:08 Guest54 let's say minetest 5.1 15:08 Guest54 again, bullying is not something i like 15:09 Guest54 (unless, of course, if it is consensual) 15:09 Guest54 appguru i just want the thing where it is not racy in minetest 5.5+ 15:09 appguru yeah, doing it right now 15:10 Guest54 thank you 15:10 appguru and not kicking < 5.3 clients ^^ 15:10 Guest54 and if you feel like it, maybe ensure the map is only sent to whoever sees is in range and not everyone 15:10 Guest54 that does constitute an info leak and also packet spam 15:12 Guest54 hmm, mesh reconstruction sure seems slow :/ 15:12 Guest54 i have a noticeable delay sometimes between placing a node and seeing it 15:12 Guest54 was it always like that? 15:12 Guest54 like it is sub-second, but weird 15:12 appguru btw 15:13 appguru should these maps be cached on the client (non-ephemeral)? 15:13 appguru Guest54: I think sending only to people in range isn't really possible 15:13 appguru That would require delaying sending the entity until you've had a chance to send them the media 15:13 Guest54 appguru yes they should 15:14 Guest54 i don't understand, please explain 15:14 Guest54 like what is the scenario here 15:15 Guest54 hmm, repixture pigs really can't swim and die to make food instead 15:16 appguru hmm lol I think I just realized something, I have to remove players from the to_receive table if they leave 15:16 appguru Guest54: As soon as you use dynamic media, all clients that might ever get in contact with it must have received it. 15:17 appguru Scenario: You do not sent dynamic media to a client. Later on, the server sends the entity using the dynamic media you have not sent (yet). You now try to send the dynamic media but it is too late, the entity arrives before the media. 15:17 appguru did not send* 15:17 Guest54 wait can't you change the texture on an entity later on 15:17 Guest54 or am i mistaken about that 15:18 Guest54 or respawn the entity? 15:18 appguru Yes, you could redistribute the media to all clients in proximity each time a new client gets in proximity. 15:18 Guest54 like, send dynamic media, respawn it, or is it too late? 15:18 Guest54 how do *you* do it? 15:18 Guest54 like in your own mods 15:18 Guest54 i am glad you have thought about it 15:18 Guest54 usually people who have solved problems before have deeper insights (sometimes not, ofc) 15:18 appguru I must bite the bullet and send to all players 15:19 Guest54 for what in particular in your mods? skins? 15:21 appguru skins, yes 15:21 Guest54 appguru you can do the “send to all players” thing ofc, if you don't have the motivation to work it out. after all, i am doing that right now. 15:21 Guest54 but please file an issue then 15:21 Guest54 regarding the info leak ;) 15:21 appguru well, the info leak merely is a map 15:21 Guest54 merely 15:21 Guest54 it can betray who is where 15:22 Guest54 and what areas were mapped 15:22 Guest54 if you have the seed of the server you can figure out where people are 15:22 Guest54 if they have natural terrain 15:22 appguru it's not as bad as straight up coords which are leaked easily 15:22 Guest54 oh yeah, fun times 15:22 Guest54 particle spammers need to be sent to only people in range to prevent that 15:22 appguru Guest54: well, maps don't need to be placed as entities, right? 15:23 appguru most of the time they'll just be used by the player wielding them, right? 15:23 Guest54 uh 15:23 Guest54 they can be placed on the side of blocks 15:23 Guest54 as an upright sprite 15:23 appguru so you could simply send to that player 15:23 Guest54 which is the main purpose actually 15:23 appguru and only send to all players when it is placed 15:24 Guest54 oh i see, but how would it work when someone drops the map and someone else gets it? 15:24 MinetestBot 02[git] 04SmallJoker -> 03minetest/minetest: Particle spawner: Move definitions to source 13f8e0778 https://github.com/minetest/minetest/commit/f8e0778fc9e402674b091fcf529d0295acd4ddd0 (152023-04-08T15:23:17Z) 15:24 MinetestBot 02[git] 04SmallJoker -> 03minetest/minetest: Particle spawner: Fix missing parameters 13ba2fee0 https://github.com/minetest/minetest/commit/ba2fee07518d9c6154799b61ba1b14a2cb43ace7 (152023-04-08T15:23:17Z) 15:24 Guest54 regarding particle spammers, some time ago, some clown used the mcl2 particles that were created when running to detect players 15:24 appguru Guest54: The map would send the media to the new wielder as soon as it is wielded. 15:24 Guest54 and griefed them 15:25 Guest54 appguru sounds like a good idea. does that approach work for all versions down to 5.3? 15:25 appguru I don't think 5.3 allows selectively sending to players 15:25 appguru Before the param table, you had to send to all players I think? 15:26 Guest54 appguru i think so. well then just send it to a single player if it is supported ig 15:27 Guest54 some ubuntu LTS version is on minetest 5.4 and i don't want to cut them off 15:27 Guest54 also some mineclone5 player went to great lengths to make it run on 5.3 15:27 Guest54 i don't even know why she did that 15:27 Guest54 but she had a bunch of polyfills or so 15:27 Guest54 (that was when 5.5 was out already) 15:28 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Disable ubuntu-18.04 CI workflows 137bfea9b https://github.com/minetest/minetest/commit/7bfea9b0fe811f0356d3a61fc2a1adb72db190b7 (152023-04-08T15:27:28Z) 15:53 mrkubax10 how can I define on_use for item while also preserving default behavior (i.e. mining for example)? 15:53 Krock mrkubax10: you can't 15:53 Krock use register_on_punchnode or dignode 15:54 mrkubax10 alright 15:56 Guest54 mrkubax10 look into the autogroup mod of mcl2 to see what is done 15:58 MinetestBot 02[git] 04srifqi -> 03minetest/minetest: Translated using Weblate (Indonesian) 134083fcf https://github.com/minetest/minetest/commit/4083fcf93a6136de5a753b7eab5f7fac3a1f8937 (152023-04-08T15:37:30Z) 15:58 MinetestBot 02[git] 04BreadT -> 03minetest/minetest: Translated using Weblate (Japanese) 1387b3e9c https://github.com/minetest/minetest/commit/87b3e9cc5f790c243897f4f5d15978c2177c41ba (152023-04-08T15:37:30Z) 15:58 MinetestBot 02[git] 04mnh48 -> 03minetest/minetest: Translated using Weblate (Malay) 13a3a4640 https://github.com/minetest/minetest/commit/a3a4640257fed7f51f1320c3141ebef7553eeb94 (152023-04-08T15:37:30Z) 15:58 MinetestBot 02[git] 04mnh48 -> 03minetest/minetest: Translated using Weblate (Malay (Jawi)) 13aa28084 https://github.com/minetest/minetest/commit/aa2808460487f972386d52c4e12cb3b4eb92e0d3 (152023-04-08T15:37:30Z) 15:58 MinetestBot 02[git] (16 newer commits not shown) 16:01 mrkubax10 Guest54: I will check, thanks 16:11 appguru Guest54: Could xmaps be leaking memory? lol 16:11 appguru (It's minor, but you don't seem to be cleaning up in on_leaveplayer) 16:13 MinetestBot 02[git] 04Wuzzy2 -> 03minetest/minetest_game: Update locale files 1334c1d32 https://github.com/minetest/minetest_game/commit/34c1d3242d3f965d9044d0e2de07e8c6cb810f88 (152023-04-08T16:11:37Z) 16:13 MinetestBot 02[git] 04Wuzzy2 -> 03minetest/minetest_game: Update German translation 13446e56a https://github.com/minetest/minetest_game/commit/446e56a879df61f3825cd7723f85819eab368326 (152023-04-08T16:11:37Z) 16:13 MinetestBot 02[git] 04appgurueu -> 03minetest/minetest_game: Fix typo in german translation 134c6e199 https://github.com/minetest/minetest_game/commit/4c6e19968ac846f4b935838e83ab38cc173c1e6f (152023-04-08T16:11:37Z) 16:15 MinetestBot 02[git] 04appgurueu -> 03minetest/minetest_game: Validate & sanitize formspec fields (#3022) 13833ed77 https://github.com/minetest/minetest_game/commit/833ed776204b194d96b50e400861dec07b5c16aa (152023-04-08T16:13:45Z) 16:39 appguru Guest54: btw another consideration is that a malicious or broken client could easily block usage of maps for everyone (by refusing to acknowledge that the media has been loaded) 16:39 Guest54 appguru then what should be done? 16:49 sfan5 apparently /usr/bin/file knows how to identify Minetest schematics, cool 16:49 appguru Guest54: I don't know. Timeout? 16:50 Guest54 appguru sounds good 17:01 appguru argh my Thunderbird is drunk again 17:13 muurkha sfan5: you know you've made it in life when /etc/magic lists your file type 17:31 sfan5 !op 17:31 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Bump version to 5.7.0 131b95998 https://github.com/minetest/minetest/commit/1b95998d11fb0d51fbfeedc56de4ca2191ad3f7a (152023-04-08T16:04:52Z) 17:31 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Continue with 5.8.0-dev 13c2a9ac2 https://github.com/minetest/minetest/commit/c2a9ac24ac2e74b111b4bbdf42ee3a62514bc7a2 (152023-04-08T16:05:03Z) 17:55 sfan5 Minetest 5.7.0 has been released! read about the changes or grab it here: https://forum.minetest.net/viewtopic.php?f=18&t=29356 18:03 Yad Ooh, that's big news! ^^ 18:18 MinetestBot 02[git] 04numberZero -> 03minetest/minetest: Remove fast faces (#13216) 1335929d2 https://github.com/minetest/minetest/commit/35929d27e3a93085c3a27180c1805711dcfe95d5 (152023-04-08T18:17:15Z) 18:18 MinetestBot 02[git] 04Desour -> 03minetest/minetest: Get rid of wgettext 1367068cf https://github.com/minetest/minetest/commit/67068cfaf43ef95e526401d9f788790516b9f8ed (152023-04-08T18:17:50Z) 18:21 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Move video_driver default selection to runtime 13c26e122 https://github.com/minetest/minetest/commit/c26e122485a7180edf434c0c1211713ff377d6d7 (152023-04-08T18:19:35Z) 18:21 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Drop ENABLE_GLES option 139d736e8 https://github.com/minetest/minetest/commit/9d736e8b8baeeacad9cfa94edd18adfcaf000029 (152023-04-08T18:19:35Z) 18:24 independent56 build fails with XInput not found 18:24 independent56 Despite the fact i don't use xbox 18:25 Desour are you building with correct irrlichtmt? 18:25 Desour (also, does xbox use X11?) 18:25 rubenwardy xinput is DirectX's input API 18:26 rubenwardy well not sure about directx, I assume that's where the x comes from, it's a Windows input API though 18:26 sfan5 xinput is a thing with X11 18:27 sfan5 install libxi-dev 18:27 Desour is it the same xinput though? 18:27 rubenwardy oh that's dumb, there's also X Input 18:28 independent56 i dunno 18:28 independent56 i'm getting the minetest/irrilichtmt thing 18:29 independent56 I installed it and no longer get the error 18:32 independent56 ChatGPT said it was for xbox 18:32 rubenwardy maybe don't trust chatgpt 18:33 independent56 yeah 18:33 independent56 It gets you half the way there and drops you off a cliff 18:35 Guest54 lol 18:42 independent56 Ithink i'm getting there 18:42 independent56 Minetest 7 here we come 18:42 independent56 *5.7 18:58 Guest54 so one of the recent commit seems to have messed with rendering, there are now way more triangles. you can see this clearly in wireframe view. can someone explain to me why this is a good thing? 18:59 Guest54 (i am assuming it was deliberate) 19:01 Desour fast faces was removed. the number of triangles is not what makes minetest rendering slow 19:01 Guest54 Desour so was fast faces slow lol? 19:03 Guest54 i thought the most efficient way to render stuff was to combine faces, so i wonder why this was a) introduced b) removed 19:05 Desour it was not slow, but not significantly faster either 19:05 Guest54 was the PR tested on a lower-spec machine? 19:06 Desour idk 19:06 Guest54 i mean spamming a drawcall for each particle is “not slow” on whatever gaming machine devs have 19:06 Guest54 but you can lag out machines ppl use for minetest by just spamming particles, if they are too low-spec 19:07 Guest54 (the solution is to batch drawcalls or use the irrlicht particle system, but the latter was removed from irrlichtmt and i don't know how to do the former) 19:08 rubenwardy spamming a drawcall is slow especially on gaming machines 19:08 rubenwardy fast faces doesn't effect the number of drawcalls - it's the same with or without 19:08 rubenwardy it only effects mesh sizes and time to make the mesh 19:08 Guest54 yeah i was just wondering what it affects 19:09 Guest54 like there must be a reason it was added and i wonder why 19:15 Guest54 celeron55_ it seems you added fastfaces in the beginning, what is the significance of more or less triangles? 19:20 rubenwardy that would have been done in 2011 ish when celeron55 was using an ancient laptop 19:20 rubenwardy the technique was probably effective for devices from early 2000 which we no longer support 19:20 rubenwardy *2000s 19:21 Guest54 no longer support? 19:21 Guest54 minetest so far ran fine on anything that supports at least opengl 1.4 (even without shaders) and GL extensions from 2003 forwards 19:22 Guest54 the loading bar looks weird on anything pre-2003 kinda 19:22 Guest54 well, i'll test it on my personal machine ig 19:22 Guest54 (am not on it rn) 19:34 Guest54 rubenwardy well i just tested it on a very fast machine i have access to and removing fast faces is definitely an fps drop, it's just not noticeable if your fps is limited anyway. 19:35 Guest54 you can test that yourself as well, just look at flat ground in devtest 19:36 Guest54 since there exist machines that get like 20 to 25fps (which is playable) with minetest, i have no good hopes for them 19:36 Desour "just tested on a very fast machine" "fps is limited anyway" "look at flat ground in devtest" these don't fit together 19:37 Guest54 Desour what do you mean? 19:37 Desour by what means is that machine fast? 19:38 Guest54 i limited fps to 600 instead of 60 and the machine is a thinkpad p14s, which is about the fastest machine i have access to (but do not own) 19:39 Guest54 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Cezanne (rev d1) (prog-if 00 [VGA controller]) 19:40 Guest54 Desour i got a drawtime of 2 to 3ms looking at a flat plane with view range 300 with fast faces. i get 3 to 4ms looking at the same thing without fast faces. 19:40 Desour ah, so you're complaining because you get 500 instead of 600 fps in a super simple scene? 19:40 Guest54 well, what i am implying is that anything that got 25fps before will probably not be so lucky 19:41 Desour no, you can't conclude that 19:41 Guest54 then i should test it 19:41 Desour it's not like new hardware just makes everthing by a factor faster :P 19:41 Guest54 also it was about “the technique was probably effective for devices from early 2000 which we no longer support” 19:41 Guest54 i mean, it definitely affects newer devices 19:42 Guest54 just not in a way anyone cares 19:42 Guest54 unless you up the triangle count a ton lol 19:42 rubenwardy this was tested on older devices, according to the PR 19:43 Guest54 well, i will test it myself 19:43 Guest54 later 19:43 Guest54 i just don't understand why it was done in the first place 19:43 Guest54 if it is so unproblematic to remove it 20:16 Guest54 rubenwardy you might want to make an announcement post here: https://old.reddit.com/r/Minetest/ 20:16 rubenwardy waiting for blog post 20:17 Guest54 oh, sorry then! 22:34 Desour huh, github look so ... small 22:34 Guest54 are you tripping or what 22:35 Guest54 DPI adjustment? 22:35 Guest54 that being said, i have Xft.dpi: 284 for one high res display (3840x2160) and minetest thinks it's still 72 lol 22:36 Desour I didn't change my setup 22:36 Guest54 maybe accidental zoom? 22:36 Guest54 ctrl + 0 22:36 Guest54 resets zoom 22:36 Guest54 in browsers 22:37 Guest54 uh, so i now set screen_dpi = 284 in minetest.conf and i don't think it did anything? 22:38 Guest54 ok, gui_scaling = 2.5 does the trick 22:38 Guest54 but i don't get it, what's the DPI value for then 22:39 Desour screen_dpi is only used as fallback. use dpi_scaling (or whatever it's called) 22:40 Guest54 well, every other app i tried does the correct thing from the Xft.dpi setting, maybe it just needs to be read 22:40 Guest54 btw is the screen size stuff going to clash badly with client size gui scaling and HUD scaling? 22:41 Guest54 like if i have a large screen and have hud_scaling set to 2.5, are mods going to get the info that they should send me giant formspecs, or is that taken into account? 22:42 Desour mods can't directly access those settings. they're just used for transforming formpsec coordinates and similar to pixels 22:43 Guest54 the question is if the thing where the mod can know your screen size is going to show your real screensize 22:43 Guest54 or if it's clever like CSS pixels 22:43 Guest54 where a “pixel” is basically some ideal thing at arms length 22:43 Guest54 not a physical pixel 22:44 Desour oh, you mean that newly added screensize API. idk how that works 22:44 Guest54 1 px = 1/96th of 1 CSS inch 22:44 Guest54 Desour well if it works like i think it works it may clash, but i may be a bit of a dummy 22:50 rubenwardy Guest54: DPI is basically meaningless, even the OS will lie about it 22:51 Guest54 rubenwardy well every application except some games seem to respect xft.dpi for text rendering, at least under linux 22:52 Guest54 rubenwardy anyways, since you did the feature where you can know the window size, what happens with people who have hud scaling on for obv reasons, will that clash with the server knowing their window size? 22:52 Guest54 i mean i *think* it is meant for scaling up formspecs server-side right? 22:52 rubenwardy the server is also sent the hud_scaling * dpi / 96 22:52 rubenwardy which is the same multiplier used inside the engine 22:53 Guest54 which would mean the value would need to be adjusted for hud_scaling if used for hud scaling 22:53 Guest54 good 22:53 Guest54 thanks! 22:53 rubenwardy gui_scaling is used for formspecs 22:53 Guest54 well i have both 22:53 Guest54 gui_scaling works for the main menu 22:53 Guest54 hud_scaling works for, well, hud elements 22:53 Guest54 xmaps map for example 22:54 Guest54 i wonder if i should make the xmap scale based on this 22:54 Guest54 hey lizzy 23:08 muurkha yeah, a CSS px is a unit of angle