02:45 < bridge> <sans._.> Does the server send info about everyone connection to everyone or just players nearby? Say something like a player hundreds of tiles away, would the client receive info about them like position, angle, and other effects that might be applied
02:45 < bridge> <sans._.> Does the server send info about everyone connected to everyone or just players nearby? Say something like a player hundreds of tiles away, would the client receive info about them like position, angle, and other effects that might be applied
02:45 < bridge> <sans._.> Does the server send info about everyone connected to everyone or just players nearby? Say something like a player hundreds of tiles away, would the client receive info about them like position, angle, and other effects that might be applied?
02:46 < bridge> <louis> yes if you have showall ejabled
02:46 < bridge> <sans._.> What about headless clients like swarfey's lib?
02:47 < bridge> <louis> idk if it's changed but default is showall is disabled unless the client sends /showall
02:47 < bridge> <sans._.> Alright
02:49 < bridge> <sans._.> So with this in mind, couldn't a bot join a server, monitor everyone and run it's own calculations to figure out who's cheating? Kinda like how antibot does it?
02:50 < bridge> <louis> yeah you'd get all the information but probably not as much as antibot
02:50 < bridge> <louis> idk if you get cursor position unless you're specifically spectating a character
02:50 < bridge> <sans._.> I mean what else is there
02:50 < bridge> <louis> only angle
02:51 < bridge> <sans._.> Does ab use the cursor position?
02:51 < bridge> <louis> im not sure though, if u check the spectator cursor pr it'll probably show u
02:51 < bridge> <louis> yes cursor position is sent to servers so im sure antibot uses it
02:51 < bridge> <sans._.> This is worth tinkering with tbh
02:52 < bridge> <louis> also u probably dont get as much network info as server gets, for example how many packets were dropped or whatever
02:52 < bridge> <sans._.> probably yeah, but still good to know
02:53 < bridge> <sans._.> I wanna try to train an AI of some sort, based on the info a normal mod would get
02:53 < bridge> <louis> fun
02:53 < bridge> <sans._.> and have it spit out confidence values for each player
02:53 < bridge> <sans._.> 
02:53 < bridge> <sans._.> i.e. how confident is it that a certain player is cheating
02:54 < bridge> <louis> r u training with antibot output as labels?
02:54 < bridge> <sans._.> I have 0 knowledge of how to train AI and I needed an idea
02:54 < bridge> <sans._.> so... idk?
02:54 < bridge> <louis> ah
02:55 < bridge> <sans._.> Am trying to learn fun things
02:55 < bridge> <sans._.> :owo:
02:55 < bridge> <louis> yeah its a good idea, but you need a ground truth data if u want to do supervised training
02:56 < bridge> <louis> or maybe u can predict the # of points a tee has based on their inputs lmao
02:56 < bridge> <louis> u can get the label data a lot easier for that
02:57 < bridge> <sans._.> I mean, I don't see why I couldn't make a quick script to monitor and store anonimyzed tee data like position changes and angle and whatnot
02:57 < bridge> <sans._.> Would just fill a lot of space real quick
02:58 < bridge> <sans._.> Also isn't score in race gamemodes just their finish time in number form?
02:58 < bridge> <sans._.> like seconds or smth
03:27 < bridge> <totar> I've made client side AB before, it's a lot harder with only 25tps instead of the servers full 50tps. It takes much more data to get the same confidence and even still it's not as accurate
07:46 < bridge> <louis> can someone explain
07:46 < bridge> <louis> ```c++
07:46 < bridge> <louis> if(m_NumObjectsHit < 10)                          m_apHitObjects[m_NumObjectsHit++] = pChr;
07:46 < bridge> <louis> ```
07:46 < bridge> <louis> in handle ninja? if more than 10 objects are hit does it just not track
07:55 < bridge> <louis> seems like a rly bad way to track if an object was hit before or not
09:23 < bridge> <louis> https://cdn.discordapp.com/attachments/293493549758939136/1382983814569132032/targetswitch_2025-06-13_00-23-13.mp4?ex=684d238b&is=684bd20b&hm=4ee081ab7612dd13f06f405b7b08999830456578cd839abb9d257a5b1c8e8388&
09:27 < bridge> <louis> btw apparently this actually happens and tees will get 'hit' by ninja more than  once if they're the 11th and onwards tee to get hit
09:28 < bridge> <louis> at least visually seems like that's the case
09:28 < bridge> <louis> https://cdn.discordapp.com/attachments/293493549758939136/1382985075808927825/Nucl_2025-06-13_00-26-26.mp4?ex=684d24b8&is=684bd338&hm=53da0500d24843d4a6641c409d118a35e6490dfbbbb1bd3b8291a2de7af72257&
09:29 < bridge> <louis> (in the vid only the tees near the end of the slash get sent super high)
09:33 < bridge> <inv41idu53rn4m3> tees are liquid
10:03 < bridge> <Jupstar ✪> Tees are atoms
10:04 < bridge> <Assa> you can do incredibly cool stuff by adding a quadart and then setting a texture for it
10:04 < bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1382994091360387102/screenshot_2025-06-13_10-03-58.png?ex=684d2d1e&is=684bdb9e&hm=610ee7895569dcb35d1b0055eba0e896ca600f2e108d8348fd9efb4e36b6bcb6&
10:09 < bridge> <Jupstar ✪> Debian
11:07 < bridge> <ryozuki> make gentoo logo
11:50 < bridge> <cyberfrighter> cool
12:15 < bridge> <zhn> wat
12:16 < bridge> <zhn> oh seems my collectables draft becomes useless
12:19 < bridge> <zhn> based on server, some servers don't have /showall just for sake of bandwidth. if sv_high_bandwidth enabled, servers sends info on 50 tps rate otherwise only 25 tps. so if both conditions are met, you can get the same data server has, but quantized (i don't remember if server quantizes data on its end too, probably yes)
12:20 < bridge> <sans._.> how would you know if any given server has high bandwidth enabled?
12:20 < bridge> <sans._.> or /showall
12:20 < bridge> <zhn> ctrl + shift + d, check recv num on the top left corner. /showall is just a chat command so you have to prompt it
12:21 < bridge> <sans._.> ah i thought some servers have it disabled
12:21 < bridge> <zhn> afair 0xf and f ddrace with 128 players have some limitations on this
12:22 < bridge> <sans._.> it'll only concern me when it comes to normal ddrace
12:22 < bridge> <sans._.> sooo
12:22 < bridge> <zhn> only ddnet block servers are 25 tps
12:22 < bridge> <zhn> and all servers should have /showall enabled
12:22 < bridge> <zhn> not by default though
12:52 < bridge> <Solly (bun bun)> finally, circles in ddnet
12:52 < bridge> <Solly (bun bun)> oh wait
13:25 < bridge> <Jupstar ✪> The oldschool map Last Chance has 176 animations (envelopes)
13:25 < bridge> <Jupstar ✪> 
13:25 < bridge> <Jupstar ✪> And it uses literally none of it xD
13:53 < bridge> <zhn> i wonder if its data lost or just troll mapper
13:54 < bridge> <zhn> data loss*
13:58 < bridge> <blaiszephyr> how do i link against an older version of a lib in cmake?
13:58 < bridge> <blaiszephyr> 
13:58 < bridge> <blaiszephyr> trying to compile something that requires an older version of openssl
13:59 < bridge> <blaiszephyr> how do i link against an older version of a lib in cmake?
13:59 < bridge> <blaiszephyr> all solutions seem to be hacky
13:59 < bridge> <blaiszephyr> 
13:59 < bridge> <blaiszephyr> trying to compile something that requires an older version of openssl
13:59 < bridge> <Jupstar ✪> I bet he opened some giga map, removed all layers and started his map xD
14:00 < bridge> <Jupstar ✪> You can include openssl as sub directory (have a sub module for your git to the ssl git).
14:00 < bridge> <Jupstar ✪> 
14:00 < bridge> <Jupstar ✪> Not sure how well it works tho
14:01 < bridge> <Jupstar ✪> If pre-built is an option you can also ship the correct version of openssl and link against that
14:02 < bridge> <blaiszephyr> i see, thanks!
14:03 < bridge> <zhn> is there any other map with more than >150 envelopes from heinrichs archive? (used ofc xd)
14:03 < bridge> <Jupstar ✪> teetown has many
14:17 < bridge> <Jupstar ✪> The oldschool map `ddrace_loop` has two front layers.
14:17 < bridge> <Jupstar ✪> 
14:17 < bridge> <Jupstar ✪> @robyt3 that is actually something that should potentially be validated and rejected. The editor considers the later layer as the front layer while ignoring the first.
14:17 < bridge> <Jupstar ✪> But it's quite possible other logic would do the opposite.
14:19 < bridge> <Jupstar ✪> @patiga Is the above a hard error in twmap-check?
14:29 < bridge> <patiga> no, also appeared too often, its a WARN log iirc
14:30 < bridge> <patiga> yea, some mappers like that ^^
14:30 < bridge> <patiga> https://discord.com/channels/252358080522747904/746534464984973323/795982789651464192
14:30 < bridge> <Jupstar ✪> I meant that a map has two front layers
14:30 < bridge> <Jupstar ✪> or generally game layers duplicated
14:31 < bridge> <patiga> yep, I understood it like that
14:31 < bridge> <Jupstar ✪> I only found 1 map yet
14:31 < bridge> <Jupstar ✪> But IMO this is a rather critical error
14:31 < bridge> <patiga> hm, might've softened that from an error after going over heinrich's archive at some point
14:31 < bridge> <Jupstar ✪> It potentially changes gameplay if parsing changes
14:32 < bridge> <patiga> yea, parsing mustn't change I determined ^^
14:33 < bridge> <patiga> honestly my error/warn policy with heinrich's archive is inconsistent at this point
14:33 < bridge> <patiga> e.g.I introduced tilemaps with non-divisible by 16 textures as an error
14:34 < bridge> <Jupstar ✪> This also affects rendering components
14:34 < bridge> <patiga> wdym?
14:35 < bridge> <Jupstar ✪> The rendering component will simply assume it's a second front layer
14:35 < bridge> <patiga> interesting, twmap simply completely dismisses the dupes
14:35 < bridge> <Jupstar ✪> In which order?
14:35 < bridge> <Jupstar ✪> The later is the real front layer?
14:35 < bridge> <patiga> I'd need to look it up, I think the last layer is the correct one
14:36 < bridge> <Jupstar ✪> Anyway, we should fix that for official maps
14:36 < bridge> <patiga> I'd be fine with making this an error, yea
14:36 < bridge> <Jupstar ✪> I don't see any reason to keep bugs
14:36 < bridge> <Jupstar ✪> Idc about the archive, that is so many unused and broken maps anyway xD
14:37 < bridge> <patiga> on my to-do list is a fix-map function operating on the datafile, it should try to make as many maps as possible pass, but be rough and inefficient to not become the norm ^^
14:38 < bridge> <Jupstar ✪> Yes sounds generally good
14:38 < bridge> <patiga> I do care about the archive because I want, yk, a map lib you can use for any teeworlds/ddnet maps
14:39 < bridge> <patiga> which parser are you working on, ddnet-rs's?
14:39 < bridge> <Jupstar ✪> Yes I have no problem if you want a auto fix thing.
14:39 < bridge> <Jupstar ✪> 
14:39 < bridge> <Jupstar ✪> It's more about when a converter can decide it on it's own.
14:39 < bridge> <Jupstar ✪> And in this case IMO the converter can not
14:39 < bridge> <Jupstar ✪> I already have a converter, but I was checking all ddnet maps and saw some glitches in few maps (mostly oldschool)
14:40 < bridge> <patiga> yea, I already fixed a couple of stuff in older ddnet maps. other stuff I decided to work around in twmap
14:42 < bridge> <Jupstar ✪> The unused env map I only found bcs it found a way to spawn more characters in a `to_lossy` call from rust than there were utf8 characters in the src string xD
14:42 < bridge> <Jupstar ✪> And since the parser only passes if it can convert back & forth it lost a few bytes in the end of the string
14:52 < bridge> <patiga> those are the cases where twmap throws a warning
14:53 < bridge> <Jupstar ✪> For names I also simply do lossy calls. Like I care if some character is lost in an invalid name.
14:54 < bridge> <Jupstar ✪> Which does not mean I wouldn't fix them for official maps tho 😬
15:03 < bridge> <patiga> leaving them as-is allows future parsers to potentially be smarter and reconstruct the original names ^^
15:05 < bridge> <Jupstar ✪> smarter than rust's stdlib? 😄
15:05 < bridge> <patiga> I could imagine that, yes
15:05 < bridge> <Jupstar ✪> U mean like look in a directory?
15:05 < bridge> <Jupstar ✪> dictionary*
15:06 < bridge> <Jupstar ✪> Honestly official maps can also be fixed by hand (list all strings and decide if they need a manual fix or auto fix)
15:07 < bridge> <patiga> maybe, I'd bet it would be possible to recover russian names for example, as they should consist mostly of non-ascii letters
15:07 < bridge> <patiga> do the extendes ascii tables always use all slots?
15:07 < bridge> <patiga> *extended
15:07 < bridge> <patiga> @jupeyy_keks as you appear to miss my #mapping posts, here is my favorite one :)
15:07 < bridge> <patiga> https://discord.com/channels/252358080522747904/746534464984973323/1238111472702193736
15:08 < bridge> <robyt3> I guess it should be a warning and all but one front layer should be discarded (either the first or last depending on how server/client currently handle this)
15:08 < bridge> <Jupstar ✪> XDD
15:09 < bridge> <robyt3> https://xkcd.com/1738/
15:10 < bridge> <Jupstar ✪> Man robyte had that tab open
15:21 < bridge> <blaiszephyr> Sometimes you just want to be prepared :kek:
16:20 < bridge> <Assa> Robyt, the master of reviews :deen_star:
16:27 < bridge> <ryozuki> :owo:
16:27 < bridge> <ryozuki> @essigautomat what u working on? seen u rly active  lately
16:27 < bridge> <TsFreeeeeeeeeddie> love you
16:28 < bridge> <ryozuki> :poggers2:
16:29 < bridge> <ryozuki> when u have to benchmark something that lasts >20mins
16:29 < bridge> <ryozuki> and u need to rerun to know if it improves
16:29 < bridge> <ryozuki> :feelsbadman:
16:30 < bridge> <ryozuki> @milkeeycat btw ill be working on a SML compiler
16:30 < bridge> <milkeeycat> when you have to make a github workflow which takes 30mins and you have to rerun each time when it didn't work :feelsbadman:
16:30 < bridge> <ryozuki> @milkeeycat https://www.roc-lang.org/
16:30 < bridge> <ryozuki> new lang everyday
16:33 < bridge> <milkeeycat> @ryozuki im working on instruction selection(it should take generic instructions and spit out arch specific instructions) for my compiler backend
16:33 < bridge> <ryozuki> epyc
16:33 < bridge> <ryozuki> https://tenor.com/view/burger-eating-frieren-frieren-beyond-journey%27s-end-sousou-no-frieren-gif-13425073513713719938
16:33 < bridge> <ryozuki> @jupeyy_keks https://perf.rust-lang.org/dashboard.html
16:34 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1383092045996097757/image.png?ex=684d8858&is=684c36d8&hm=6853e4f33afd3222c8d803b1ab552d3cbfedb23da799bfba3134f1fa75972b8a&
16:34 < bridge> <ryozuki> rust actually is improving comp time
16:34 < bridge> <blaiszephyr> can i forcefully use bundled libs
16:34 < bridge> <blaiszephyr> can i forcefully use bundled libs in cmake
16:34 < bridge> <blaiszephyr> can i forcefully use bundled libs in cmake somehow
16:34 < bridge> <Assa> last thing I did was #10340 which almost nobody noticed, but is like a big boost to rendering performance
16:34 < bridge> <DDNet> https://github.com/ddnet/ddnet/pull/10340
16:34 < bridge> <Assa> last thing I did was #10340 which almost nobody noticed, but is like a HUGE boost to rendering performance
16:35 < bridge> <ryozuki> noice
16:35 < bridge> <ryozuki> iirc it prefers them over system if present
16:35 < bridge> <ryozuki> did u checkout submodules
16:36 < bridge> <blaiszephyr> yeah i did, i also trierd using `-DPREFER_BUNDLED_LIBS`
16:36 < bridge> <blaiszephyr> 
16:36 < bridge> <blaiszephyr> but it doesnt.. tried to use a bundled version of openSSL
16:36 < bridge> <blaiszephyr> 
16:36 < bridge> <blaiszephyr> i assume because of pkg_config?
16:36 < bridge> <blaiszephyr> https://cdn.discordapp.com/attachments/293493549758939136/1383092602731106326/image.png?ex=684d88dd&is=684c375d&hm=552b75c6d65bf33e3183163f32d0e8c1b285a715ea56e356a4148fc78f46a9c8&
16:40 < bridge> <ryozuki> r u on nix
16:40 < bridge> <ryozuki> idk ur nix shenanigans
16:40 < bridge> <ryozuki> xd
16:40 < bridge> <ryozuki> isnt it a meme the weird ld paths on nix
16:40 < bridge> <ryozuki> :justatest:
16:41 < bridge> <Jupstar ✪> Cool!
16:41 < bridge> <Jupstar ✪> 
16:41 < bridge> <Jupstar ✪> I want this for ddnet 😄
16:41 < bridge> <Jupstar ✪> I love stats like dese
16:42 < bridge> <ryozuki> yea
16:42 < bridge> <ryozuki> hard to automate cuz graphics
16:44 < bridge> <blaiszephyr> i use arch btw
16:44 < bridge> <blaiszephyr> :(
16:45 < bridge> <blaiszephyr> wouldnt be a biggie if i could just downgrade my openssl version but..
16:45 < bridge> <blaiszephyr> https://cdn.discordapp.com/attachments/293493549758939136/1383094929542348922/image.png?ex=684d8b07&is=684c3987&hm=199e65bdb1aa179b845a49020a3e0972443c7cc516157286dcf27c2756d795e1&
16:45 < bridge> <TsFreeeeeeeeeddie> there's a chinese saying "have a foot in both camps"
16:45 < bridge> <blaiszephyr> i tripleboot windoof, arch and nix :kek:
16:46 < bridge> <blaiszephyr> nix for fun, arch as my daily driver and windoof for league and a handful of work related software
16:46 < bridge> <TsFreeeeeeeeeddie> https://cdn.discordapp.com/attachments/293493549758939136/1383095165282943107/image.png?ex=684d8b40&is=684c39c0&hm=d7560e99d66d427960c3b4a30f68bd8355769fe572767adbf52369ce554c0990&
16:47 < bridge> <Assa> I'll install ubuntu on my machine in october, when the windoofs license runs out. On Laptop/at work I use ubuntu
16:47 < bridge> <Assa> I'll install ubuntu on my machine in october, when the windoofs goes EOL. On Laptop/at work I use ubuntu
16:48 < bridge> <Assa> hmm I could do crazy stuff, crontab every morning and git pull ddnet upstream and build it
16:48 < bridge> <Assa> hmm I could do crazy stuff, crontab every morning and git pull ddnet upstream and build it, so make my own nightlies :3
16:54 < bridge> <Assa> or I simply go on steam and select nightly under betas
17:14 < bridge> <Jupstar ✪> Race maps often have a second tele layer
17:15 < bridge> <Jupstar ✪> Append abuse
17:15 < bridge> <Jupstar ✪> So you need that old ssl for something ddnet related?
17:15 < bridge> <blaiszephyr> kind of, its rather teeworlds related
17:16 < bridge> <Jupstar ✪> do you have a build library for that ssl version?
17:17 < bridge> <blaiszephyr> i dont have a libssl.so if thats what you mean
17:18 < bridge> <Jupstar ✪> ok, i mean you could just clone openssl, checkout a version u need and build your things
17:18 < bridge> <Jupstar ✪> there should then be a pkg-config file in the build dir and you can somehow force cmake to find that one
17:18 < bridge> <Jupstar ✪> or hardcode the ssl_library and include paths
17:28 < bridge> <robyt3> Any idea about this log message with `dbg_gfx 4`? Happens with and without 10340.
17:28 < bridge> <robyt3> `gfx: [PERFORMANCE] (importance: medium) Program/shader state performance warning: Vertex shader in program 18 is being recompiled based on GL state.`
17:29 < bridge> <Assa> huh I never got this one?
17:29 < bridge> <Jupstar ✪> which one is program 18? Xd
17:29 < bridge> <robyt3> Happens on Launch with OpenGL 3.3
17:30 < bridge> <robyt3> With PR 10340 it's program 20 instead
17:30 < bridge> <Jupstar ✪> Can you look it up in renderdocs or dbg msg the shader id
17:30 < bridge> <Jupstar ✪> Or let me see, maybe i have the same IDs
17:30 < bridge> <Assa> yes 10340 adds 2 programms
17:30 < bridge> <Assa> yes 10340 adds 2 programms (for ogl3.3)
17:30 < bridge> <robyt3> I guess it's just the index of the program
17:31 < bridge> <Jupstar ✪> I'd assume it's matching `TWGLuint m_ProgramId;`
17:34 < bridge> <Assa> I count only 18 shaders in 10340 who are getting added by ogl, I might be missing something
17:35 < bridge> <Assa> or is it double for vertex and fragment shader?
17:36 < bridge> <Jupstar ✪> For me it's most likely `m_pSpriteProgramMultiple`
17:36 < bridge> <Assa> then it would be the textured quad fragment shader, but I assume they're in sequence (and maybe I misscount)
17:36 < bridge> <Jupstar ✪> But I dont get any warning like dis
17:37 < bridge> <Assa> me neither
17:40 < bridge> <Assa> on windows I have to run the configure script before building, so that the shader gets recompiled
17:40 < bridge> <Assa> on windows I have to run the configure script before building, so that the shader gets recompiled (at least for vulkan), because the shader compilation step is part of the build setup
17:42 < bridge> <robyt3> I tried reconfigure to make sure but still same result
17:48 < bridge> <robyt3> Program 20 should be the use of`UseProgram` in `CCommandProcessorFragment_OpenGL3_3::Cmd_RenderQuadLayer`
17:49 < bridge> <robyt3> ~~Program 20 should be the use of`UseProgram` in `CCommandProcessorFragment_OpenGL3_3::Cmd_RenderQuadLayer`~~
17:56 < bridge> <robyt3> Seems like it comes from the `UseProgram` for `shader/spritemulti.vert/frag`
17:59 < bridge> <Jupstar ✪> I dunno it makes no sense to me, I found this in the internet, which would make sense:
17:59 < bridge> <Jupstar ✪> 
17:59 < bridge> <Jupstar ✪> > My vertex shader contains some inputs like layout (location=0) in vec3 VertexPosition. But the corresponding attribute wasn't enabled yet in the state machine. So the driver compiles a program version for a state where no Attribute is enabled. Now, if I bind my first VAO for drawing, all the necessary attributes get enabled and the driver detects a mismatch between the state of my first draw call and the state when I compiled the shader. So h
17:59 < bridge> <Jupstar ✪> 
17:59 < bridge> <Jupstar ✪> But not for this shader. We always use all input attributes
18:00 < bridge> <Jupstar ✪> > I've run across this problem recently. It occurred that unbinding Vertex Array Objects caused this warning to appear. After deleting the lines with glBindVertexArray(0) the warning message disappeared.
18:00 < bridge> <Jupstar ✪> Maybe it's as stupid as that
18:01 < bridge> <robyt3> That seems to do it
18:28 < bridge> <ryozuki> https://github.com/cyphar/paperback
18:32 < bridge> <louis> how did i miss this loo
18:32 < bridge> <louis> how did i miss this lol
18:39 < bridge> <Assa> @robyt3 > The new shader programs are not being deleted yet.
18:39 < bridge> <Assa> 
18:39 < bridge> <Assa> ~~? I have no idea what this means~~ oh memory leak
18:40 < bridge> <robyt3> Search for `m_pQuadProgram`, the deletes are missing for the new programs
18:41 < bridge> <robyt3> And `DeleteProgram` calls
18:41 < bridge> <Assa> yes already doing exactly that
18:42 < bridge> <Assa> I guess valgrind can't find that, since it's in the graphics backend
18:44 < bridge> <Assa> :deen_star:
19:06 < bridge> <louis> why not coexist  i want more interactable entities
19:18 < bridge> <teero777> im getting in a physics fight with cmake
19:18 < bridge> <teero777> im getting in a physical fight with cmake
19:22 < bridge> <Ewan> why
19:24 < bridge> <Jupstar ✪> How does `ColorRGBA` in our code base know that it has to initialize to `(1.0, 1.0, 1.0, 1.0)`
19:24 < bridge> <Jupstar ✪> 
19:24 < bridge> <Jupstar ✪> I cannot find where we set a default constructor to do so
19:26 < bridge> <Jupstar ✪> @learath2 explain that sorcery :c
19:27 < bridge> <Jupstar ✪> Or does it simply use 0.0 as default
19:27 < bridge> <Jupstar ✪> Well probably lol
19:27 < bridge> <Jupstar ✪> Dunno why I even assumed it's 1.0 xd
19:27 < bridge> <Jupstar ✪> Probs bcs in colors it's often like that
19:29 < bridge> <inv41idu53rn4m3> what? Black is totally default
19:29 < bridge> <Jupstar ✪> @essigautomat see that as request changes not as approved xD
19:29 < bridge> <inv41idu53rn4m3> though yeah I prefer (0, 0, 0, 1)
19:29 < bridge> <Jupstar ✪> I had approved selected but then saw the one thing xD
19:29 < bridge> <TsFreeeeeeeeeddie> I think the default is 0,0,0,0
19:29 < bridge> <teero777> i have 1 library at the top level with a cmake file. and 2 cmake files for the tests. If i try to compile the library and the tests with -flto, the linker tells me:
19:29 < bridge> <teero777> ```bash
19:30 < bridge> <teero777> [ 46%] Linking C executable benchmark
19:30 < bridge> <teero777> cd /home/teero/software/ddnet_physics_c/build/tests/optimized && /sbin/cmake -E cmake_link_script CMakeFiles/benchmark.dir/link.txt --verbose=1
19:30 < bridge> <teero777> /sbin/clang -Xlinker --dependency-file=CMakeFiles/benchmark.dir/link.d CMakeFiles/benchmark.dir/benchmark.c.o -o benchmark  -Wl,-rpath,/home/teero/software/ddnet_physics_c/build/libs/ddnet_map_loader ../../libddnet_physics.a ../../libs/ddnet_map_loader/libddnet_map_loader.so -lm -lz -lomp
19:30 < bridge> <teero777> CMakeFiles/benchmark.dir/benchmark.c.o: file not recognized: file format not recognized
19:30 < bridge> <teero777> clang: error: linker command failed with exit code 1 (use -v to see invocation)
19:30 < bridge> <teero777> ```
19:30 < bridge> <inv41idu53rn4m3> now I'm reconsidering
19:30 < bridge> <inv41idu53rn4m3> maybe transparent black is the bestest
19:30 < bridge> <Jupstar ✪> really?
19:30 < bridge> <Jupstar ✪> I'd say white is the logical default for colors
19:30 < bridge> <Jupstar ✪> Since often stuff is textured and multiply with white => texture
19:30 < bridge> <inv41idu53rn4m3> If you like paper
19:30 < bridge> <inv41idu53rn4m3> oh true
19:31 < bridge> <inv41idu53rn4m3> ig black is default clear colour and white is default fill colour
19:31 < bridge> <TsFreeeeeeeeeddie> I prefer not assuming anything and just freaking use the four arguments constructor
19:31 < bridge> <TsFreeeeeeeeeddie> :nouis:
19:31 < bridge> <Ewan> I see `constexpr ColorRGBA(){}`
19:31 < bridge> <Jupstar ✪> Yeah ig it's simply 0.0, was too lazy to test it
19:32 < bridge> <Jupstar ✪> value initialization of floats
19:32 < bridge> <Jupstar ✪> Smart boi
19:32 < bridge> <Ewan> i think the defaults are explicitly made here
19:32 < bridge> <Ewan> https://cdn.discordapp.com/attachments/293493549758939136/1383136971387568201/image.png?ex=684db22f&is=684c60af&hm=8f16d7ef6e528b2da34fda68e00b22eeb26b6d74b2531daf9fd702ed0acfbf80&
19:32 < bridge> <Jupstar ✪> yeah
19:33 < bridge> <Ewan> :brownbear: another day of ppl blaming cmake for linker issues
19:34 < bridge> <inv41idu53rn4m3> xyza is crazy
19:34 < bridge> <inv41idu53rn4m3> though I guess we can't really pull the glsl magic trick where r and x are the same thing
19:34 < bridge> <TsFreeeeeeeeeddie> It is
19:34 < bridge> <TsFreeeeeeeeeddie> It's a union type
19:35 < bridge> <TsFreeeeeeeeeddie> The default construcor just decided to use xyza
19:36 < bridge> <Ewan> :brownbear:
19:36 < bridge> <Ewan> https://cdn.discordapp.com/attachments/293493549758939136/1383137849133629551/image.png?ex=684db300&is=684c6180&hm=fb9cc5aaee4ba8a998a6b9bce991525448dea9a138b92d03e425d2f39bdfc023&
19:36 < bridge> <Ewan> it's got a few constructors
19:42 < bridge> <Ewan> btw using LTO takes away any ABI stability you had in generating the object files
19:43 < bridge> <Ewan> LTO is really dependent on the compiler's internal representation so you can't expect to use it unless your compiler + linker are on the same page about it
19:43 < bridge> <Ewan> meaning, clang must use ldd to link with its LTO, and they have to be the same versions
19:43 < bridge> <Ewan> -fuse-ld=ldd
19:44 < bridge> <Ewan> but also... probably don't use -flto directly
19:44 < bridge> <Ewan> because cmake has settings for it which wrap compilers
19:44 < bridge> <Ewan> called IPO
19:44 < bridge> <Ewan> `set_property(TARGET name-target-here PROPERTY INTERPROCEDURAL_OPTIMIZATION TRUE)`
19:45 < bridge> <Ewan> or for all targets `set(CMAKE_INTERPROCEDURAL_OPTIMIZATION TRUE)`
19:45 < bridge> <Ewan> if you want to check it then you can use this <https://cmake.org/cmake/help/latest/module/CheckIPOSupported.html#module:CheckIPOSupported>
19:46 < bridge> <Ewan> i'd recommend doing so but it won't rly matter as it just moves the error to generate time instead of link time lol
19:46 < bridge> <Ewan> also `cmake_minimum_required(VERSION 3.9)` would be correct to add but optional
19:46 < bridge> <Ewan> it's a fairly new feature
19:55 < bridge> <inv41idu53rn4m3> wew
19:55 < bridge> <inv41idu53rn4m3> You're lacking uvpq
19:55 < bridge> <inv41idu53rn4m3> oh wait v is taken
19:55 < bridge> <Jupstar ✪> rip
19:56 < bridge> <Jupstar ✪> isn't it uvrt anyway?
19:56 < bridge> <Jupstar ✪> never heard of pq, but maybe i dunno
19:56 < bridge> <inv41idu53rn4m3> I'm not familiar with rt 😳
19:57 < bridge> <inv41idu53rn4m3> oh there's stpq too
19:58 < bridge> <inv41idu53rn4m3> rt overlaps with rgb
19:58 < bridge> <inv41idu53rn4m3> I need to look up the spec now lol
20:00 < bridge> <Jupstar ✪> st for uv, my brain would take a while xD
20:00 < bridge> <Jupstar ✪> just remove HSV support
20:02 < bridge> <inv41idu53rn4m3> Okay GLSL supports xyzw, rgba, and stpq
20:02 < bridge> <inv41idu53rn4m3> uv was my own hallucination
20:06 < bridge> <inv41idu53rn4m3> man can be a pain looking up glsl documentation lol
20:06 < bridge> <inv41idu53rn4m3> it's so scattered
20:06 < bridge> <inv41idu53rn4m3> no I'm not insane enough to read the spec
20:19 < bridge> <teero777> i won the fight against cmake
20:21 < bridge> <Ewan> how did you do it
20:21 < bridge> <blaiszephyr> 0.50€ he switched to bam
20:27 < bridge> <teero777> somehow i was really stupid. i was using target_compile_options instead of target_link_options which messed things up. but after i fixed that, more errors appeared. then i upgraded to clang20 which fixed those errors. but the llvm20-libs are not on the AUR or official arch linux repos smh so i downgraded again and then magically worked.
20:27 < bridge> <teero777> somehow i was really stupid. i was using target_compile_options instead of target_link_options which messed things up. but after i fixed that, more errors appeared. then i upgraded to clang20 which fixed those errors. but the llvm20-libs are not on the AUR or official arch linux repos smh so i downgraded again and then it magically worked.
20:27 < bridge> <teero777> somehow i was really stupid. i was using target_compile_options instead of target_link_options which messed things up. but after i fixed that, more errors appeared. then i upgraded to clang20 which fixed those errors. but the llvm20-libs are not on the AUR or official arch linux repos  so i was missing libs for the sanitizers.so i downgraded again and then it magically worked.
20:27 < bridge> <Ewan> target compile options can influence the linker behavior but it depends on the flag
20:27 < bridge> <teero777> somehow i was really stupid. i was using target_compile_options instead of target_link_options which messed things up. but after i fixed that, more errors appeared. then i upgraded to clang20 which fixed those errors. but the llvm20-libs are not on the AUR or official arch linux repos  so i was missing libs for the sanitizers. so i downgraded again and then it magically worked.
20:28 < bridge> <Ewan> but yea
20:28 < bridge> <Ewan> glad it works for you ig
20:28 < bridge> <Ewan> using compiler-specific flags when an abstracted version exists is still kinda pointless but if it works, it works
20:32 < bridge> <teero777> im already using a bunch of clang specific things in my codebase so trying to make flags work with other compilers would be a waste of time
20:36 < bridge> <Ewan> ~~in your case using cmake is already a waste of time~~
20:43 < bridge> <jxsl13> -in your case
20:43 < bridge> <jxsl13> 👀
20:48 < bridge> <Ewan> there are lots of great reasons to use cmake
20:48 < bridge> <Ewan> but the entire two step build system thing is completely useless if you only use one compiler xd
20:48 < bridge> <Ewan> and that is the main reason why it exists
20:48 < bridge> <Ewan> or, the reason most projects use it
21:08 < bridge> <zhn> cuz same logic can be replicated
21:13 < bridge> <louis> ya ig
21:13 < bridge> <louis> i wanna try my hand at server side dummies too at some point
21:22 < bridge> <teero777> @mpft cmake is more readable than just a makefile imo. that is one of the reason i stick to it over just a makefile
21:24 < bridge> <teero777> @mpft cmake is more readable than just a makefile imo. that is one of the reason i stick to it over makefiles
22:36 < bridge> <patiga> twmap-py update: macos on arm should now also be supported. (also they are actually cross-compiled, not native builds like I originally thought)