00:11 <+bridge> yo is everything down rn? 00:11 <+bridge> ? 00:12 <+bridge> Cloudflare is not down 00:20 <+bridge> slopbot fixing the slop made by the other slopbot was not on my 2026 bingo 00:20 <+bridge> https://cdn.discordapp.com/attachments/293493549758939136/1504971666218811402/image.png?ex=6a08ed96&is=6a079c16&hm=0b078f73b472ff52fd2d27795fd201a84248911fd4cc82a74ca664878f774343& 00:22 <+bridge> It goes even further, another 2 slopbots commented on the slopbots PR, and they are having a slopchat in the PR going back and forth refining the slop 00:22 <+bridge> Future of software engineering 00:34 <+bridge> Damn, what an absolute mess that is. Bun shouldn't have pushed it into main, absolutely, but that wasn't even released yet and people are already storming out of using bun. They are mad at internal development as if it was already released. 00:34 <+bridge> 00:34 <+bridge> I use bun for literally all my projects, but at least I will wait and only get mad if it's something that affects released versions. lol 00:35 <+bridge> It's very obvious some of those people commenting are hardcore anti-AI developers. Most of them don't even know what's happening and are only there for the drama with off-topic comments. lol 00:35 <+bridge> Bun was clearly in too deep before this, ive never used it before but I definitely wont be using it now. 00:36 <+bridge> yeah my projects use Bun too, but this is just.. i dont want rust as a dependency in my go projects, for example 00:36 <+bridge> The only thing I was already against Bun was that they keep releasing features without fixing previous ones, but it wasn't that bad. 00:37 <+bridge> yeah my projects use Bun too, but this is just.. i dont want rust as a dependency in my go projects, for example :ong: 00:37 <+bridge> I just found the amount of AI layers involved in this mess just hilarious. Don't have much more opinion on what Bun was, but definitely won't be trusting the AI slop rewrite of it with anything 00:37 <+bridge> i love it, actually 00:37 <+bridge> you have AI agents complain about AI agents 00:37 <+bridge> Honestly, if it works and the releases have less bugs, I don't really care how it was made. That's a nightmare for another team to manage, not my problem. xd 00:38 <+bridge> Yeah, basically. xd 00:39 <+bridge> deno already exists, and is written in rust. Presumably with less slop 00:39 <+bridge> Deno just wasn't inovative enough. 00:40 <+bridge> i just want cargo for all my existing projects pls ty 00:40 <+bridge> preferably a bit faster, but overall i want dis 00:41 <+bridge> Bun's interfaces and features are so well thought. Most likely implemented by AI, but it's obvious that the human behind them knows what he's doing. I believe people just have to trust the process. 00:42 <+bridge> Bun had to maintain their own version of Zig before, which i find super funny 00:43 <+bridge> I do not trust clankers 00:43 <+bridge> but you do you, the monied interests surely agree more with your stance than mine, so you might be on to something 00:43 <+bridge> but you do you, the monied interests surely agree more with your stance than mine, so you might be onto something 00:45 <+bridge> The zig repo banned all ai assisted contributions, so I think they gave up hope on getting their fixes upstreamed 00:47 <+bridge> You shouldn't even trust yourself, let alone something that is only meant to be used as a tool. I don't think anyone should expect AI to work by itself, there are "can make mistakes" labels literally everywhere. 00:47 <+bridge> 00:47 <+bridge> I don't think this was a problem of AI, it was a management problem. They just shouldn't have pushed it to main, it makes no sense. 00:48 <+bridge> You shouldn't even trust yourself, let alone something that is only meant to be used as a tool. I don't think anyone should expect AI to work by itself, and besides, there are "can make mistakes" labels literally everywhere. 00:48 <+bridge> 00:48 <+bridge> I don't think this was a problem of AI, it was a management problem. They just shouldn't have pushed it to main, it makes no sense. 00:49 <+bridge> In the worse case, the AI rewrite would have failed and they would give up on the branch. 00:54 <+bridge> In the worse case, the AI rewrite would have failed and they would just give up on the branch without affecting anything else other than time. 00:55 <+bridge> In the worse case, the AI rewrite would have failed and they would just give up on the branch without affecting anything other than time used to try. 00:57 <+bridge> they didn't even put the rust code is a separate folder. all the source files are currently sitting next to each other 00:57 <+bridge> https://cdn.discordapp.com/attachments/293493549758939136/1504980954802815016/image.png?ex=6a08f63c&is=6a07a4bc&hm=3a4f572a08c5f09a9ed3019b2c6d24936a0baee76894de426e31b55c0a6b69c1& 00:58 <+bridge> every rust file has dozens of unsafe 💀 01:00 <+bridge> Definitely a rushed push. I can only imagine the big men upstairs at Anthropic gave them a small deadline, and they just went with it. lol 01:02 <+bridge> I can't imagine anthropic had anything to do with this 01:02 <+bridge> they're not stupid 01:04 <+bridge> I would hope not. They are not stupid, but I'm just not sure how much they actually care about Bun's stability. Anthropic gets to test their bots and improvement with a test in battle, and Bun is just the battle field. 01:05 <+bridge> I would hope not. They are not stupid, but I'm just not sure how much they actually care about Bun's stability. Anthropic gets to test their bots and improvement with a test in battle, and Bun might just be the battle field. 01:06 <+bridge> I would hope not. They are not stupid, but I'm just not sure how much they actually care about Bun's stability. Anthropic gets to test their bots and to improve them with a test in battle, and Bun might just be the battle field. 07:25 <+bridge> ## itsabot_rust 07:25 <+bridge> - a version of `itsabot`, with custom sounds! 07:25 <+bridge> - credit to eNJi for the skin! 07:25 <+bridge> 07:25 <+bridge> ## scifi-basic 07:25 <+bridge> 07:25 <+bridge> - Show off your futuristic weapons! Custom looks and sounds! 07:25 <+bridge> - Credit to @broso56 for the weapon skins 07:25 <+bridge> - Credit to Tristis for the audio! 07:25 <+bridge> 07:25 <+bridge> https://cdn.discordapp.com/attachments/1278298263841407009/1321969818751991961/fun_times_indeed_2_small.mp4?ex=67751a95&is=6773c915&hm=a99f56da44bd5271a31f3773d2c8f5c6ac622d1eda7ba26f7f1ec30f2663a4c7& 07:25 <+bridge> https://cdn.discordapp.com/attachments/293493549758939136/1323688481964429343/itsabot_rust.7z?ex=6a094337&is=6a07f1b7&hm=49461551f97c0deb8da279f0c46736feb6d1e7376ee71ca84db6aebbaf445792& 07:25 <+bridge> https://cdn.discordapp.com/attachments/293493549758939136/1323688482320941117/sci-fi_basic.7z?ex=6a094337&is=6a07f1b7&hm=16c66170a559eb7a7a47ca55f485a1b3e9e74ac885cdf1da2adcc1c77f6b7264& 07:27 <+bridge> @drafakiller: oh interesting. So you want to do some client sided moderation bot for ddnet? Cool! I would argue that tracking competitive block stats and detecting griefing blockers in a race are slightly different. I wrote that code forever ago but it is probably close to what is in ddnet++ and F-DDrace which is the type of block tracking I personally like. So the last toucher counts as killer. Meaning kills can be stolen. I think even while 07:28 <+bridge> @drafakiller: i allow a short time I forgot how long maybe 2 seconds to pass between last interaction and the victim becoming frozen. If they die on their own in that time it still counts as a kill. But the timer is usually long enough that long falling times should not fail to track the kill. 07:31 <+bridge> @drafakiller: there is some chat command check !help which prints every single kill as soon as it happens it can show you more details. Or check a block server running ddnet++, S-DDrace, F-DDrace to see it in the kill feed. (Would love to name the block gametype here too, but Heinrich broke that) 07:34 <+bridge> @drafakiller: I would track griefing entirely different btw. Because if you apply my logic on a regular ddrace run it will always detect block kills as soon as a team fails in a part and respawns. 07:34 <+bridge> 07:34 <+bridge> I would focus on the spawn because that is the biggest problem anyways. Since mappers fail to optimize against spawn blockers most of the time. So here you have to limit the area in which you track kills. And I would try to catch intention instead of just pure amount of touches leading to kills. And I would define spawnblock intention as someone pulling another tee back with the hook that is trying to walk into the other direction. 08:15 <+bridge> :omo: 09:05 <+bridge> @learath2 hi 09:06 <+bridge> how was that bpf guy called 09:06 <+bridge> new kernel got something cool 09:06 <+bridge> bpf based schedulers 09:06 <+bridge> im gonna use this https://crates.io/crates/scx_lavd 09:06 <+bridge> SCHED_CLASS_EXT on kernel config 09:07 <+bridge> https://packages.gentoo.org/packages/sys-kernel/scx 09:07 <+bridge> https://github.com/sched-ext/scx 09:07 <+bridge> this means they are switchable at userland too 09:08 <+ChillerDragon> ryo u mean rei? 09:08 <+bridge> and one can say my scheduler is made in rust 09:08 <+bridge> yes @reitw u will love this 09:09 <+ChillerDragon> maybe davide likes it too but i think he is more C enjoyer hrhr 09:09 <+bridge> > sched_ext is a Linux kernel feature which enables implementing kernel thread schedulers in BPF and dynamically loading them. This repository contains various scheduler implementations and support utilities. 09:09 <+bridge> > 09:09 <+bridge> > sched_ext enables safe and rapid iterations of scheduler implementations, thus radically widening the scope of scheduling strategies that can be experimented with and deployed; even in massive and complex production environments. 09:09 <+bridge> bpf is lang agnostic, its just this scheduler was made in rust 09:10 <+bridge> from what i gather, scx_lavd is the best option for gaming 09:10 <+bridge> there is also scx_rusty and scx_bpfland 09:16 <+bridge> ok upon more investigation it seems scx_lavd is better for single ccd wand my cpu got 2 09:31 <+ChillerDragon> @learath2 can you look into https://github.com/ddnet/ddnet/pull/10964 please 12:02 <+bridge> Hmm, which bpf guy? 12:11 <+bridge> Hm, sure, I guess that's as good a thing as any to do 12:33 <+bridge> ReiTW 12:33 <+bridge> apparently this was possible with out of tree patches since 2024 12:33 <+bridge> but i guess they added it now 13:53 <+bridge> yeah that's pretty sick lol https://docs.kernel.org/scheduler/sched-ext.html 13:55 <+bridge> oh wow, would love to see some benchmarks regarding this 15:27 <+bridge> yes when the slot is not active, or when data mismatches (e.g. name) 15:27 <+bridge> https://cdn.discordapp.com/attachments/293493549758939136/1443668159738544329/screenshot_2025-11-27_19-21-31.png?ex=6a095e7f&is=6a080cff&hm=468d3103d8a9ccaf4c79aded2a4360beba3d6fd2ebaf028a3d64f4799e094b65& 18:40 <+bridge> aww `webm` doesnt work embedded? 18:55 <+bridge> Try adding empty lines above and below the attachment link 19:05 <+bridge> what a weird thing, thanks xd 19:41 <+bridge> # check my bio 😁 22:10 <+bridge> this would be insanely useful for moderation 22:14 <+bridge> For sure, I just keep opening the demos on a 2nd client. 22:14 <+bridge> Someone gave me a very practical suggestion which was having the client launch a second client to view the demo, its not ideal but its much better than opening a second client and finding the demo manually 22:27 <+bridge> In Firefox, the Emscripten client could support viewing a live demo stream :justatest: 22:28 <+bridge> @brokecdx you could use my template to technically make this for tpl so you can watch people play directly from the website. 22:30 <+bridge> dope 22:37 <+bridge> Do you use the JS functions to write the demo files? That might work with live demos in any browser. But Emscripten natively does only support streaming files in Firefox currently. Did you see #12160? Should allow skins etc. to be downloaded. 22:37 <+bridge> https://github.com/ddnet/ddnet/pull/12160 22:38 <+bridge> nice, that would allow me to majorly reduce the download size since i have to include all the skins right now 22:39 <+bridge> yea