04:04 <+bridge_> <12944qwerty> hm, if areas are small, which is usually, then it will look weird
04:04 <+bridge_> <12944qwerty> https://cdn.discordapp.com/attachments/293493549758939136/1473877763986427985/image.png?ex=6997cf22&is=69967da2&hm=798212b551aa6335cbebac35e5d11ea27f04c71e40d5e55639c5a310686bfac7&
04:04 <+bridge_> <12944qwerty> what magic number to use to know what size to cutoff boundary vs centroid hmmm
04:05 <+bridge_> <12944qwerty> also, the colors are using the ddteamcolors
04:07 <+bridge_> <12944qwerty> also, the colors are using the ddteamcolors, and it's not great
06:50 <+bridge_> <gorp_tw> oh im trolling it's barely moved
06:51 <+bridge_> <gorp_tw> but hm, i would really prefer if the maximum dimension is over a certain large threshold, something like 20 tiles
06:51 <+bridge_> <gorp_tw> centroid is probably fine for many regions
06:52 <+bridge_> <gorp_tw> Tunelock are tiles that set player-specific tunes, regardless of where they move
06:52 <+bridge_> <gorp_tw> it's a pending PR, needs a map
06:53 <+bridge_> <gorp_tw> this seems to not follow what i suggested, i think you should focus on a narrow list of very distinct colors that repeat after like 16 or 32 maybe
06:53 <+bridge_> <gorp_tw> and based on number of distinct inclusions rather than hardcoded mapping of tune number -> color
07:32 <+bridge_> <gorp> does anyone find it funny that Matrix has this totally not-deen picture under every one of his commits
07:32 <+bridge_> <gorp> like who is this guy
07:33 <+bridge_> <gorp> ive thought about it for MONTHS now and it was a small little funny for myself to enjoy but i need other people to know about it
08:01 <+bridge_> <12944qwerty> I had done it the same way that team colors were done, hsl and golden ratio
08:01 <+bridge_> <12944qwerty> Didnt do anything else
08:02 <+bridge_> <12944qwerty> The code sucks rn tho, when i zoom out, i get 5fps lol
14:23 <+bridge_> <Assa> > I am okay with merging. But I rather have someone else press the button 😛
14:23 <+bridge_> <Assa> @chillerdragon I could press the button, leave it unreviewed and you still keep all the blame. Do you have arthritis or are you just whiny?
14:24 <+bridge_> <Assa> I think you should not be afraid to do a mistake, mistakes happen, even if you review the code, check the PR out, test it ingame, write a din a4 page discussion, ...
14:36 <+bridge_> <Assa> Instead say "I am unsure about these changes, I'd like to have another maintainer take a second look" and in theory just add another reviewer. But asking us to press the button for you is something of an unding
15:25 <+bridge_> <fomjn> @learath2 can I get purple?
15:25 <+bridge_> <fomjn> @essigautomat ty for helping with my GH questions :heartw:
15:36 <+bridge_> <manki233> I want to build DDNet with mingw and ninja, but it got wrong with it:
15:36 <+bridge_> <manki233> https://cdn.discordapp.com/attachments/293493549758939136/1474052025791680554/fea6452f8ec02cdac9a139791dd6a357.png?ex=6998716d&is=69971fed&hm=20a44c744bcaf2cbaf9693e554738a73c5d377c3f4602a973ef8f57b161b4e0f&
15:37 <+bridge_> <ryozuki> https://www.sandordargo.com/blog/2026/02/18/cpp26-std_is_within_lifetime
15:37 <+bridge_> <ryozuki> @learath2 https://clang.llvm.org/docs/BoundsSafety.html
15:38 <+bridge_> <manki233> I can build the lib without error, but it only spawn .rlib files
15:38 <+bridge_> <manki233> https://cdn.discordapp.com/attachments/293493549758939136/1474052469477736559/3004719ae5108e64ae35d14556dcb453.png?ex=699871d7&is=69972057&hm=44c3aa61f11092e72e62730bc2b67d013c9921d74460ae97b45b6664b10f34ba&
15:40 <+bridge_> <manki233> What  can I deal with it?
15:40 <+bridge_> <ryozuki> https://nicolasdickenmann.com/blog/the-great-fp64-divide.html
15:41 <+bridge_> <ryozuki> u shouldnt do anything with this, its used as an internal compile step
15:41 <+bridge_> <fomjn> https://discord.com/channels/252358080522747904/1125706766999629854/1147096614238957621
15:41 <+bridge_> <fomjn> Need to (edited) the message
15:42 <+bridge_> <fomjn> Add to 「Achievement Roles」:
15:42 <+bridge_> <fomjn> "@Mapper: Assigned to users whose map was released."
15:49 <+bridge_> <robyt3> Set the correct rust toolchain for mingw for your build folder
15:49 <+bridge_> <robyt3> Set the correct rust toolchain for mingw/gnu for your build folder
15:49 <+bridge_> <blaiszephyr> we really need to see if we can detect this in cmake
15:50 <+bridge_> <blaiszephyr> This is probably the most annoying wall people hit xd
15:50 <+bridge_> <robyt3> Yeah, it's also not explained in the readme it seems
15:51 <+bridge_> <blaiszephyr> I thought I added something to it.. I'll check later
15:51 <+bridge_> <robyt3> Open command prompt in your build folder, then run `rustup override set stable-x86_64-pc-windows-gnu`
15:53 <+bridge_> <manki233> I will try
16:00 <+bridge_> <manki233> ok, I probably clear...
16:00 <+bridge_> <manki233> https://cdn.discordapp.com/attachments/293493549758939136/1474058078839242896/image.png?ex=69987710&is=69972590&hm=13429ac2c16a0260d9b6cb68f0a6b9e3b36163307b8201b9bcb590fc9aa2205c&
16:03 <+bridge_> <robyt3> You need gnu, not msvc
16:04 <+bridge_> <robyt3> The gnu toolchain should be active for your gnu/mingw-based build folder, you can have the msvc toolchain installed at the same time for other build folders
16:06 <+bridge_> <alw5> @furo do you still have that list of the most popular skins?
16:11 <+bridge_> <manki233> yes. Thanks!
20:47 <+bridge_> <Assa> I currently let AI run over my border clipping code, it is calculating for half an hour and is still unsure wtf I am doing :kek:
20:51 <+bridge_> <milkeeycat> I thought about it more and I think I'll go with: HIR -> gets lowered into gMIR -> instruction selection replaces generic instructions with target specific ones(MIR). I think it's easier to do/reason about instruction selection on lower level instructions, and I basically ended up with the same stuff I already have xdd. 
20:51 <+bridge_> <milkeeycat> 
20:51 <+bridge_> <milkeeycat> HIR - high level IR, has abstract values, structs, maybe other high-level stuff.
20:51 <+bridge_> <milkeeycat> gMIR - generic MIR, some of the instructions will be the same as in HIR(add instruction for example), but it will also have more low level ones, and also support use of target specific instructions. This IR will work with vregs instead of abstract values, and it won't have structs.
20:51 <+bridge_> <milkeeycat> MIR - target specific instructions only.
20:53 <+bridge_> <h1se2> https://cdn.discordapp.com/attachments/293493549758939136/1474131806134603860/image.jpg?ex=6998bbba&is=69976a3a&hm=237d12ba4e1071e40c9c3cacc964334af909af7d513d897d7c789ccb05aa573c&
20:53 <+bridge_> <h1se2> https://cdn.discordapp.com/attachments/293493549758939136/1474131806944231464/image.jpg?ex=6998bbba&is=69976a3a&hm=f0d2d9bf17d6c7ad7f3acbaf3ecb2a37d0b55caf79c01f7f919928549aec7a99&
20:53 <+bridge_> <h1se2> https://cdn.discordapp.com/attachments/293493549758939136/1474131807703142420/image.jpg?ex=6998bbbb&is=69976a3b&hm=4942dee62216427dde27d0df3193d578814563ae64d46ef3bed2c5ca7fd619be&
20:53 <+bridge_> <h1se2> https://cdn.discordapp.com/attachments/293493549758939136/1474131808257052906/image.jpg?ex=6998bbbb&is=69976a3b&hm=9712ee592062a998981b69ff575391c98a9d97bfc04dbe51acbe5f914d849585&
20:53 <+bridge_> <milkeeycat> @Discord Mod
20:55 <+bridge_> <louis> which model
20:56 <+bridge_> <Assa> `big pickle` :kek:
21:10 <+bridge_> <Assa> it finished, it produced code that compiled - but its not working at all
21:41 <+bridge_> <Assa> Instead I asked the AI to split the PR into 2 commits now, hopefully that's better :justatest:
21:56 <+bridge_> <learath2> Have you decided what to do with structs?
21:58 <+bridge_> <milkeeycat> ye, I'll "lower" them, so if I have 2 i64 struct store, it will be represented as 2 store instructions in gMIR
21:58 <+bridge_> <milkeeycat> it works like this right now
22:00 <+bridge_> <milkeeycat> ye, I'll "lower" them, so if I have a store instruction of a struct which consists of 2 i64, it will be represented as 2 store instructions in gMIR
22:04 <+bridge_> <learath2> Nice, I like it