00:02 < bridge> The entire accounts proposal relies on how quic does encryption. The client login certificate is the client certificate for the quic connection 00:03 < bridge> (or tls, basically connection layer binding of the channel to keys) 00:31 < bridge> oh shit 00:50 < bridge> <01000111g> Does someone know, why in the ddnet datasette on db.ddstats.org, the data of the time cps is just 0 for all rows in the race table? I kinda need that info 00:52 < bridge> You sure it’s all 0? Maps that don’t have time cps have those 0d iirc 00:52 < bridge> <01000111g> No, its not all, but in the webview i also see negative values and it feels weird. I just tried to fetch all rows in the race table of Stronghold and only got 0s, but maybe i did sth wrong, will have to look deeper 00:53 < bridge> https://db.ddstats.org/ddnet-128c319?sql=select+*+from+race+where+cp1%3D21.40 00:53 < bridge> 00:53 < bridge> forks for me 00:53 < bridge> https://db.ddstats.org/ddnet-128c319?sql=select+*+from+race+where+cp1%3D21.40 00:53 < bridge> 00:53 < bridge> works for me 01:02 < bridge> <01000111g> ``` 01:02 < bridge> <01000111g> import requests 01:02 < bridge> <01000111g> import pandas as pd 01:02 < bridge> <01000111g> 01:02 < bridge> <01000111g> BASE_URL = "https://db.ddstats.org/ddnet.json" 01:02 < bridge> <01000111g> MAP_NAME = "Stronghold" 01:02 < bridge> <01000111g> BATCH_SIZE = 5000 01:02 < bridge> <01000111g> offset = 0 01:02 < bridge> <01000111g> all_rows = [] 01:02 < bridge> <01000111g> columns = None 01:02 < bridge> <01000111g> 01:02 < bridge> <01000111g> 01:02 < bridge> <01000111g> while True: 01:02 < bridge> <01000111g> sql = f""" 01:02 < bridge> <01000111g> SELECT * 01:02 < bridge> <01000111g> FROM race 01:02 < bridge> <01000111g> WHERE Map = :map 01:02 < bridge> <01000111g> ORDER BY Timestamp ASC 01:02 < bridge> <01000111g> LIMIT {BATCH_SIZE} OFFSET {offset} 01:02 < bridge> <01000111g> """ 01:02 < bridge> <01000111g> 01:02 < bridge> <01000111g> params = { 01:02 < bridge> <01000111g> "sql": sql, 01:02 < bridge> <01000111g> "map": MAP_NAME 01:02 < bridge> <01000111g> } 01:02 < bridge> <01000111g> 01:02 < bridge> <01000111g> r = requests.get(BASE_URL, params=params) 01:03 < bridge> <01000111g> r.raise_for_status() 01:03 < bridge> <01000111g> data = r.json() 01:03 < bridge> <01000111g> 01:03 < bridge> <01000111g> if columns is None: 01:03 < bridge> <01000111g> this gives me 0s and i dont get why 01:03 < bridge> hm idk, i sadly dont have the energy right now to debug xd 01:05 < bridge> hard code it 01:06 < bridge> stronghold doesnt have time cps 01:06 < bridge> https://ddstats.tw/map/Stronghold/timecps 01:06 < bridge> try a map with time cps 01:06 < bridge> <01000111g> xd 01:07 < bridge> <01000111g> how is this possible, I thought we added time cps to all maps 01:07 < bridge> nope 01:07 < bridge> <01000111g> gg 01:07 < bridge> there was a spreadsheet 01:07 < bridge> that had the data 01:07 < bridge> iirc brutal 50% done? 01:07 < bridge> <01000111g> somehow, there is a single run in Brazil on Sh1 with time cps???? 01:07 < bridge> idk tho 01:07 < bridge> what where 01:07 < bridge> <01000111g> https://db.ddstats.org/ddnet-128c319?sql=select+*+from+race+where+map+%3D+%22Stronghold%22 01:07 < bridge> <01000111g> first one 01:08 < bridge> uhh 01:08 < bridge> i have no idea 01:08 < bridge> probably a bg 01:08 < bridge> probably a bug 01:09 < bridge> <01000111g> im going to crashout soon. I need them to do data science 🙁 01:10 < bridge> none of the strongholds have time cps 01:10 < bridge> gg 01:10 < bridge> why isnt the timecp project 01:10 < bridge> based off most finishes 01:11 < bridge> what are u making 01:18 < bridge> doesn't this make it kinda unreasonable for vanilla based servers to integrate accounts? 01:18 < bridge> unless you're forking ddnet that sounds really hard 01:19 < bridge> <01000111g> TAS autodetection 01:19 < bridge> <01000111g> with semi-supervised learning probably, not rule-based 01:25 < bridge> well insta has reimplemented all vanilla gamemodes 01:25 < bridge> and more 01:25 < bridge> true 01:25 < bridge> do servers actually run insta 01:26 < bridge> not the fng and zcatch ones 01:26 < bridge> no one plays vanilla of any kind 01:26 < bridge> x-x 01:26 < bridge> but they could! 01:27 < bridge> wait why do u need timecps then 01:28 < bridge> it's not tas detection it's duplicate run detection 01:28 < bridge> do u just want to find runs w identical timecps as tas so u have labels 01:28 < bridge> yeah 01:29 < bridge> i guess u can label all duplicate runs and try to classify the teehistorian data of those versus non duplicates 01:30 < bridge> ive wanted to try but im not really confident that it'll be very accurate 01:32 < bridge> <01000111g> yeah, but i will not define this explicitly. Maybe i hand label some rule-based like that, but maybe not. I will explore first. There are other metadata one can assess and some features one can design (like relative time spent in certain areas of map in comparison to average players etc etc, i will play around a lot. Later, if one is able to define a global model, taking into account all maps and runs it will become very interesting, as sus playe 01:33 < bridge> 🫡 01:33 < bridge> <01000111g> Its cheated rank detection from metadata. It will be very complicated and data heavy if one actually wants to do it just by pure gameplay via teehistorian 01:35 < bridge> yah 01:35 < bridge> I think teehistorian is the answer tho to somehow get a good automatic detection 01:38 < bridge> <01000111g> yeah, but its a crazy load of data. I just try this for fun and see if i even get somewhere. 09:13 < bridge> @01000111g: I doubt the records db has enough data for TAS detection. You might want to have a look at teehistorian instead. 09:14 < bridge> Souly I run fng and zcatch! And the fng one also has been forked and is used by others. 09:52 < bridge> It is a hurdle, but I think the way the quic patchset is isolated from the rest of the code makes it easier to apply 09:55 < bridge> How will AI help you in this case? TAS runs should have all the same timecps, there is nothing you gain by AI-ing this, but introducing an error rate 09:56 < bridge> The timecps do not have enough information to do anything else with, there is nothing to learn there that then would apply to timecps of other players 10:25 < bridge> gumo ^.^ 10:25 < bridge> @learath2 ^ this one is very easy to review and merge 😄 10:41 < bridge> today is elf loader day 10:41 < bridge> today is saturday 10:42 < bridge> xd 10:42 < bridge> if i do a elf loader i can run some c or rust programs 10:56 < bridge> I also like elfs :justatest: 11:05 < bridge> <01000111g> That's not true. Not all TAS runs get replayed twice or more. Still one can see they are suspicous, because they have insane time/rank in comparison to player points and other metainformation/confirmed legit ranks. Of course a trained model has a FPR and other characteristics. But one can adjust for a specific direction one wants to go. In best case a model trained on all maps, players and races gives out a Likelihood of runs being cheated. Some 11:07 < bridge> I just don't believe there is enough information in there. You'll have a stupid large fpr for good players 11:07 < bridge> AI isn't magic. It can't learn patterns where there are no patterns 11:09 < bridge> (Feel free to mess around with it. It's just an educated guess that it'd be a waste of your time) 11:09 < bridge> <01000111g> Yeah, we'll see. I am aware of this, its just a fun side project. There are patterns definitely, but getting a reliable estimate is a challenge for sure. 12:14 < bridge> @learath2 do u think i can brick a pc if i try my kernel on real hardware 12:14 < bridge> :justatest: 12:14 < bridge> i should get a old pc to try stuff 12:34 < bridge> If the firmware isn't buggy, you shouldn't really be able to 12:34 < bridge> thing is, how would i install it 12:34 < bridge> ok qemu i pass a raw image 12:35 < bridge> on* 12:35 < bridge> would it work with a usb? 12:35 < bridge> it has a uefi bootloader 12:36 < bridge> i thought i should port it to use limine bootloader it has a rly nice crate 12:40 < bridge> @ryozuki i have old hardware 12:40 < bridge> 12:40 < bridge> 1050ti, ryzen5, old hdd 12:40 < bridge> 12:40 < bridge> if you want me to test it, just tell me the steps xd 12:41 < bridge> i wonder how long a CI would take that tests an entire kernel 12:57 < bridge> between a day and a week maybe? 12:58 < bridge> might also depend on the tests, which only run on specific hardware 13:07 < bridge> qemu 13:07 < bridge> kernel itself isnt heavy, in fact should be fast to test, itd the user programs 13:23 < bridge> Can i test the kernel? 13:29 < bridge> If it's an uefi executable you can just have any bootloader chain into it. Or even directly boot it 13:44 < chillerbot> Elo @ryozuki I just wanted to remind you that: implement a basic filesystem 13:45 < bridge> i hate the fact that there is not a default filesystem for all linux applications xd 13:46 < bridge> do u want to? i can provide u the img and how to run it 13:46 < bridge> dolphin, thunar, egui's own, qt's own 13:46 < bridge> but i dont want to share the source yet 13:46 < bridge> i wanna cleanup and only share it when it has useful stuff 13:46 < bridge> i'd also like to test, i'm fine with just an img 13:49 < bridge> I guess ext3/4 is about as close as it gets to a "default" 13:50 < bridge> ext5 plz doesnt care about being stopped on write and compression 13:51 < bridge> btrfs is a pain to setup# 13:51 < bridge> btrfs is a pain to setup 13:51 < bridge> hm ye, but even then you have ext3/4, some use fat32, btrfs, and that weird thing opensuse uses 13:51 < bridge> btrfs is shit 13:51 < bridge> no one uses fat32 13:51 < bridge> why 13:51 < bridge> I had no issues what so ever running or setting up btrfs 13:51 < bridge> bit rot protection sucks 13:51 < bridge> This laptop is my first experiment with it and it hasn't given me a single issue in 16 months or so now 13:51 < bridge> zfs is the better way 13:52 < bridge> Surely you mean bitrot detection 13:52 < bridge> bitrot protection would be sick 13:52 < bridge> detection works, but it's annoying 13:52 < bridge> How would you "protect" against bitrot? It happens while nothing is running 13:53 < bridge> well it was a pain the first time when iu had no time what im doing, but now its fine i benefit greatly from compression x-x 13:53 < bridge> its just annoying that u have to have another grub pkg and testdisk doesnt work on it (and the btrfs version is wack) 13:53 < bridge> the drive should run at least 👀 13:54 < bridge> ```bash 13:54 < bridge> # run with, remove --enable-kvm if u dont have kvm 13:54 < bridge> qemu-system-x86_64 --enable-kvm -serial stdio -device isa-debug-exit,iobase=0xf4,iosize=0x04 -no-reboot -usb -m 128 -drive if=pflash,format=raw,unit=0,file=out/ovmf_prebuilt/x64/code.fd,readonly=on -drive if=pflash,format=raw,unit=1,file=out/ovmf_prebuilt/x64/vars.fd -drive format=raw,file=out/edos_uefi.img 13:54 < bridge> ``` 13:54 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1403707921866555412/edos-kernel.zip?ex=68988860&is=689736e0&hm=5bafdbce409c5da9bff78dfefa51af336520cfe9c4f9673028b4cfcb86063756& 13:54 < bridge> zipbomb inc 13:54 < bridge> As far as I remember both zfs and btrfs provides bitrot detection, both using crc32 13:54 < bridge> ill do first fat16 13:54 < bridge> then fat32 13:54 < bridge> then something better 13:54 < bridge> i guess 13:54 < bridge> do ntfs :koggers: 13:55 < bridge> yeah, the bitrot detection works and with snapshots and manual intervention you can recover your files with btrfs, but that one heck of an annoyance. 13:55 < bridge> i havent made a filesystem before 13:55 < bridge> Sane idea, FAT is very easy to work with 13:55 < bridge> i wonder if you could emulate vfs and just steal the kernal modules 13:55 < bridge> How does zfs do it better? 13:55 < bridge> @blaiszephyr u cant do anything in the kernel rn, just type stuff 13:55 < bridge> it runs a program on startup 13:55 < bridge> at least to my knowledge it is able to automatically recover those files 13:55 < bridge> that prints hello 13:55 < bridge> i dont expect your kernel to magically solve for X, dont worry :D 13:56 < bridge> i think just getting it to work on real hardware is a win 13:56 < bridge> xd 13:56 < bridge> as long as ur hardware doesnt have weird bugs it should work tm 13:56 < bridge> Hm, I would very much like to see sources on that. As far as I know both just store crcs for blocks for bitrot detection, a crc isn't an ecc, it simply doesn't contain enough information to correct the error 13:56 < bridge> but its uefi only 13:57 < bridge> and I'm guessing x86_64 only? 13:57 < bridge> 13:57 < bridge> 1. cant change setup with zfs 13:57 < bridge> 2. btrfs is inhouse kernal module, zfs is not (dkms is not required) 13:57 < bridge> 3. less ram usage / energy effecient (/shrug) 13:57 < bridge> 4. per file/per volume compression 13:57 < bridge> crc is not parity, for sure. 13:57 < bridge> is only answers a single question 13:57 < bridge> whether the file is the same as when it was written 13:58 < bridge> yeah, no 32 bits either, just 64, there is so much stuff that changes, but i have plans to make a riscv kernel targeting the QEMU virt board, the thing with riscv is there is no standard boot process for all boards, each board has different stuff 13:58 < bridge> you making me rm -rf my entire disk i see xd 13:58 < bridge> Technically crc is not cryptographically secure, so there exists a non-trivial set of bit manipulations you can do on a file that doesn't change its crc 13:58 < bridge> 13:58 < bridge> 1. raid is better (sadly doesnt effect me yet) 13:58 < bridge> idk if it works but try just putting it on a usb 13:58 < bridge> So it's only really good against detecting non-malicious changes 13:58 < bridge> and then boot from the usb 13:58 < bridge> 13:58 < bridge> 1. raid is better (sadly doesnt effect me yet) on zfs 13:58 < bridge> it shoudl work tm 13:58 < bridge> 13:58 < bridge> 1. cant change setup with zfs 13:58 < bridge> 2. btrfs is inhouse kernal module, zfs is not (dkms is not required) 13:58 < bridge> 3. btrfs has less ram usage / energy effecient (/shrug) 13:58 < bridge> 4. btrfs has per file/per volume compression 13:58 < bridge> the eufi firmware is already in ur pc 13:59 < bridge> (ovmf is the uefi firmware for qemu, because qemu needs it to have "uefi") 13:59 < bridge> Is it a raw disk image that just contains an uefi binary? If so it likely won't boot 13:59 < bridge> no, it has a bootloader 13:59 < bridge> 13:59 < bridge> 1. raid is better (sadly doesnt effect me yet) on zfs 13:59 < bridge> 13:59 < bridge> 2. zfs has better default compression 14:00 < bridge> 3. lots of jargin 14:00 < bridge> The uefi firmware reads from a FAT filesystem in a specific folder by default 14:00 < bridge> im using this https://github.com/rust-osdev/bootloader 14:00 < bridge> > Both implementations load the kernel at runtime from a FAT partition. This FAT partition is created 14:00 < bridge> > The configuration is read from a special section of the kernel's ELF file, which is created by the entry_point macro of the bootloader_api library. 14:00 < bridge> @learath2 14:01 < bridge> cool cool cool 14:01 < bridge> > It provides functions to create FAT-formatted bootable disk images, based on the compiled BIOS and UEFI bootloaders. 14:01 < bridge> file out/edos_uefi.img 14:01 < bridge> out/edos_uefi.img: DOS/MBR boot sector; partition 1 : ID=0xee, start-CHS (0x0,0,2), end-CHS (0x3ff,255,63), startsector 1, 14463 sectors, extended partition table (last) 14:01 < bridge> is it actually 100% rust? 14:01 < bridge> file says this 14:01 < bridge> it is 100% rust with some hand written inline assembly 14:01 < bridge> is it actually 100% rust?, the entire kernel i mean 14:01 < bridge> ye makes sense 14:01 < bridge> cuz for example u need to write assembly for interrupt and context switches 14:02 < bridge> ooh, i thought inline assembly for performance 14:02 < bridge> even rust compiler can't stand my code :feelsbadman: 14:02 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1403709969848729600/image.png?ex=68988a48&is=689738c8&hm=912b8b975480262af7f43e8d0dddecf99b5d3c95977edd13baef833a1d78175f& 14:02 < bridge> nah, inline cuz some stuff cant be done, for example setting up syscalls and using privileged registers and ops 14:02 < bridge> like swapgs 14:03 < bridge> No there just aren't builtins to emit some assembly, nor would such assembly even fit the data model of some languages 14:04 < bridge> makes sense, this sounds interesting but i never dipped deep into these things like - ever xd 14:04 < bridge> ryo you could have your entire system run on almost 100% rust 14:04 < bridge> https://system76.com/cosmic/ 14:05 < bridge> this as a de with a rust kernel and bootloader 14:05 < bridge> ```rust 14:05 < bridge> #[allow(unused)] 14:05 < bridge> #[unsafe(naked)] 14:05 < bridge> unsafe extern "C" fn syscall_entry() { 14:05 < bridge> naked_asm!( 14:05 < bridge> // Save user stack and switch to kernel stack 14:05 < bridge> "swapgs", 14:05 < bridge> "mov gs:0, rsp", // Save user RSP to gs:0 14:05 < bridge> "mov rsp, gs:8", // Load kernel stack from gs:8 14:05 < bridge> 14:05 < bridge> // Save return info 14:05 < bridge> "push r11", // Save user RFLAGS 14:05 < bridge> "push rcx", // Save user RIP 14:05 < bridge> 14:05 < bridge> // Save syscall arguments 14:05 < bridge> "push rax", // syscall number 14:05 < bridge> "push rdi", 14:05 < bridge> "push rsi", 14:05 < bridge> "push rdx", 14:05 < bridge> "push r10", 14:05 < bridge> "push r8", 14:05 < bridge> "push r9", 14:05 < bridge> 14:05 < bridge> // Prepare args for Rust handler: 14:05 < bridge> "mov rdi, [rsp + 0x30]", // rdi = syscall_num 14:05 < bridge> "mov rsi, [rsp + 0x28]", 14:05 < bridge> "mov rdx, [rsp + 0x20]", 14:06 < bridge> "mov rcx, [rsp + 0x18]", 14:06 < bridge> "mov r8, [rsp + 0x10]", 14:06 < bridge> "mov r9, [rsp + 0x08]", 14:06 < bridge> 14:06 < bridge> my handler 14:06 < bridge> im not a pro in x86 tho 14:06 < bridge> but it works xd 14:06 < bridge> Not exactly useful information tbh, you should look into becoming a typescript expert with a quintuple certification in AWS typescript 14:06 < bridge> laughs in secret project nr. 3 # that uses typescript xd 14:06 < bridge> Oh cool, you do call into rust ASAP 14:06 < bridge> why cant i see that channel 14:06 < bridge> secret project 14:06 < bridge> ofc i love rust 14:06 < bridge> :Pepega: 14:07 < bridge> xd 14:07 < bridge> ok its a secret project that uses typescript 14:07 < bridge> either its some web thing 14:07 < bridge> or idk 14:07 < bridge> angular go brrrr 14:07 < bridge> but well if it were in rust i would be interested in knowing 14:07 < bridge> i wanted to use svelte but others disagreed 14:08 < bridge> @learath2 btw making a kernel in rust improved a lot last years, now u have a tier 2 target for kernels x86_64-unknown-none which disables red zones and stuff 14:08 < bridge> tho initially i planned on using go for that, but eh typescript is easier to do a frontend with 14:08 < bridge> before u had to make ur own target 14:11 < bridge> ```rust 14:11 < bridge> #[unsafe(naked)] 14:11 < bridge> extern "x86-interrupt" fn timer_interrupt_handler(_stack_frame: InterruptStackFrame) { 14:11 < bridge> core::arch::naked_asm!( 14:11 < bridge> // Check if we come from user mode by checking CS 14:11 < bridge> // The interrupt frame is at [rsp] 14:11 < bridge> // CS is at offset 8 14:11 < bridge> "mov rax, [rsp + 8]", // Load CS 14:11 < bridge> "test rax, 3", // Test CPL bits 14:11 < bridge> "jz 2f", // Jump if from kernel mode (CPL = 0) 14:11 < bridge> 14:11 < bridge> // Coming from user mode, swapgs 14:11 < bridge> "swapgs", 14:11 < bridge> 14:11 < bridge> "2:", 14:12 < bridge> // Save registers 14:12 < bridge> "push rax", 14:12 < bridge> "push rbx", 14:12 < bridge> "push rcx", 14:12 < bridge> "push rdx", 14:12 < bridge> "push rsi", 14:12 < bridge> "push rdi", 14:12 < bridge> "push rbp", 14:12 < bridge> "push r8", 14:12 < bridge> "push r9", 14:12 < bridge> "push r10", 14:12 < bridge> "push r11", 14:12 < bridge> "push r12", 14:12 < bridge> "push r13", 14:12 < bridge> "push r14", 14:12 < bridge> "push r15", 14:13 < bridge> i probs need to save FPU/MMX registers, SSE registers too 14:14 < bridge> or idk i dont know 14:15 < bridge> @blaiszephyr did u try the os 14:15 < bridge> busy correcting report portfolios from my apprentice rn 14:15 < bridge> ah ok i forgot ur a teacher 14:16 < bridge> also i need to setup the PC first, its in my basement xd 14:16 < bridge> xd 14:16 < bridge> ill go play some ffxiv 14:16 < bridge> new patch is out 14:16 < bridge> i'm currently waiting for events in gw2 xd 14:17 < bridge> @learath2 how many ppl in the world u think know how to make a x86_64 kernel? 14:17 < bridge> :owo: 14:17 < bridge> its actually rly fun 14:17 < bridge> its been months since i coded a lot outside work 14:18 < bridge> glad you're enjoying your time xd 14:18 < bridge> Hm, hard to estimate, but I wouldn't say there are much more than 20k. It's a very niche thing 14:19 < bridge> I included people capable of contributing to a kernel. If we are talking about making ones own, there are probably less, maybe even less than half that 14:19 < bridge> uh 14:19 < bridge> i wonder if i could contribute to kernel 14:19 < bridge> but i guess u need more knowledge about hardware and drivers 14:19 < bridge> than kernel itself 14:19 < bridge> to be useful nowadays 14:20 < bridge> torvalds kernel? its mostly maintenance and adding new hardware now anyway, right? 14:20 < bridge> less code more copy paste 14:20 < bridge> Yes, the kernel provides you a lot of structure, it's like a "framework", writing a kernel driver is not like writing a kernel 14:20 < bridge> there is another field i want to try, which is reverse engineering, but the learning stuff seems even more obscure than os dev, 14:20 < bridge> so far i dabbled into compilers and kernels, and vms 14:21 < bridge> RE is a lot of fun, i played around with it using ghdira and ddnet, as it was easier to understand what you're seeing with the source 14:25 < bridge> Having all the system.h signatures makes it a lot easier to work with teeworlds/ddnet clients 14:26 < bridge> I once started creating FLIRT signatures for all system.cpp functions to make it much faster to RE those things but then I got distracted from it 14:28 < bridge> FunctionId in Ghidra if you only need it to work in ghidra 14:39 < bridge> if you want to suffer I recommend getting into multimedia 14:43 < bridge> nah doesnt look interesting to me 14:43 < bridge> also i hate the twitter ffmpeg guy 14:43 < bridge> lol fair 15:14 < bridge> yeah the image is fine and how to run in in QEMU 15:51 < bridge> i sent it 15:51 < bridge> . 15:51 < bridge> yeah i see it now lol 15:51 < bridge> there was a bug with a cr3 page but it didnt show up in the normal kernel execution xd 15:51 < bridge> but well i fixed it now im adding elf 15:56 < bridge> okay so are there any features? 15:56 < bridge> or is it totally useless lol 15:57 < bridge> bruv i just said it 15:57 < bridge> not much yet 15:57 < bridge> but does it show in qemu? 15:57 < bridge> can u share screen 15:57 < bridge> xd 15:57 < bridge> i mean yeah it works 15:58 < bridge> nice 16:00 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1403739638518124574/PXL_20250809_135956687.jpg?ex=6898a5ea&is=6897546a&hm=a260ae2c6600410acf12e4881679dcae0744654a5a81145b31b9e2b654ae6cd4& 16:00 < bridge> i have 2 riscv boards here 16:00 < bridge> :owo: 16:00 < bridge> @pilonpl ill add a filesystem and cli as soon as i can 16:01 < bridge> do you plan on doing anything interesting? 16:02 < bridge> i mean anythin unusual 16:02 < bridge> i mean anything unusual 16:22 < bridge> I actually wonder what information you leaked about yourself with this barcode 🤔 maybe less than I think 16:24 < bridge> I actually wonder what information you leaked about yourself with this barcode 🤔 maybe less than I think (ofc not the barcode alone, but publicly available data added) 16:25 < bridge> poggies 16:25 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1403746066364498060/image.png?ex=6898abe6&is=68975a66&hm=36ea1ed35f969cdb44286df9700d0eb685cbc1a800ba622b5faaf528526d08a6& 16:27 < bridge> how does drawing to screen work without any fancy drivers 16:27 < bridge> framebuffer 16:27 < bridge> https://en.wikipedia.org/wiki/Framebuffer 16:27 < bridge> ofc your still need a driver for your graphics device 16:27 < bridge> idk how an actual gpu driver works 16:27 < bridge> magic 16:28 < bridge> yeah but this works for now xd 16:38 < bridge> The barcode is just the product id, same as the code below barcode 16:38 < bridge> I know 16:41 < bridge> fixed the newline stuff xd 16:41 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1403750042698776706/image.png?ex=6898af9a&is=68975e1a&hm=ad67a03553958bf4130d792ec170218706e4c41be3076f37cc586481d531d2b4& 16:43 < bridge> edos is the next templeos 16:44 < bridge> btw the font rendering is manual 16:44 < bridge> im using font8x8 16:44 < bridge> thats why dots have no kerning xd 16:45 < bridge> sounds like normal behavior from a bitmap font 16:46 < bridge> ye 16:51 < bridge> Nice, when doom on edos? 16:51 < bridge> :justatest: 16:53 < bridge> Btw how do you print stuff from C? 16:55 < bridge> Same procedure as every year 16:56 < bridge> `printf("your mum\n")` 16:56 < bridge> `printf("your mum is this much weight: %d\n", (int)1e10)` 16:57 < bridge> would only work if edos has the same syscall 16:58 < bridge> its probably much more simple than printf 17:03 < bridge> ```C 17:03 < bridge> static inline long syscall_3args(long syscall_num, long arg1, long arg2, long arg3) { 17:03 < bridge> long result; 17:04 < bridge> __asm__ volatile ( 17:04 < bridge> "syscall" 17:04 < bridge> : "=a"(result) 17:04 < bridge> : "a"(syscall_num), "D"(arg1), "S"(arg2), "d"(arg3) 17:04 < bridge> : "rcx", "r11", "memory" 17:04 < bridge> ); 17:04 < bridge> return result; 17:04 < bridge> } 17:04 < bridge> 17:04 < bridge> static inline long syscall_1arg(long syscall_num, long arg1) { 17:04 < bridge> long result; 17:04 < bridge> __asm__ volatile ( 17:04 < bridge> "syscall" 17:04 < bridge> : "=a"(result) 17:04 < bridge> : "a"(syscall_num), "D"(arg1) 17:04 < bridge> : "rcx", "r11", "memory" 17:04 < bridge> ); 17:04 < bridge> return result; 17:04 < bridge> } 17:04 < bridge> 17:04 < bridge> #define SYS_WRITE 1 17:04 < bridge> #define SYS_EXIT 60 17:04 < bridge> 17:04 < bridge> long write(int fd, const void* buf, long count) { 17:04 < bridge> return syscall_3args(SYS_WRITE, fd, (long)buf, count); 17:04 < bridge> } 17:04 < bridge>