01:31 <+bridge> <jxsl13> henlo delopers 🔥
08:44 <+bridge> <fokkonaut> Hi, `tw-0.7+udp://179.61.251.168:7001` broadcasts bots as it only registers on 0.7, but it's displayed in DDNet's server browser which follows 0.6 master rules and does not have an option to filter bot players. Should they be filtered on the master server already?
08:47 <+bridge> <fokkonaut> Assa falsely accused me of doing that here, while I implemented it correctly years ago already. Should go for all servers tho
08:52 <+bridge> <fokkonaut> (again, framed me  as being offfensive, @essigautomat i really dont know why you love drama so much...)
09:03 <+bridge> <bamcane_tee237> I think bots should not be displayed in the server browser as they are basically useless information.
09:05 <+bridge> <fokkonaut> https://ddnet.org/rules/master/
09:05 <+bridge> <fokkonaut> > The player count shown should accurately reflect the number of human players. The maximum player count must indicate the total number of human players that can join.
09:16 <+bridge> <Assa> Last time I checked you had 12 players, some of them listed, none  were real players. I wanted to do a bug report, I framed it very badly tbh
09:17 <+bridge> <fokkonaut> > none  were real players
09:17 <+bridge> <fokkonaut> What do you mean by that?
09:17 <+bridge> <Assa> this is already 5 months old, this will be fun to reproduce :justatest:
09:22 <+bridge> <Assa> some players like the shopkeeper were clearly bots, they died if you hook them out of their area and respawned there, other players were just, well, not playing or behaving strange, I was wondering what was up there, they might have been external bots - I checked the player browser list and the shopkeeper was in there 🤷‍♂️ this was 5 months ago, need to recheck and do a proper report
09:23 <+bridge> <fokkonaut> I think this matches your pattern of simply try to provoke drama for your own entertainment. It's impossible the shopkeeper was in browser if you tried on a 0.6 client.
09:24 <+bridge> <Assa> was on ddnet client
09:26 <+bridge> <fokkonaut> 6 years ago:
09:26 <+bridge> <fokkonaut> https://github.com/fokkonaut/F-DDrace/blame/935c834e21e72d3d6154164de8af5ed63709035f/src/engine/server/server.cpp#L2266
09:27 <+bridge> <fokkonaut> 6 years ago:
09:27 <+bridge> <fokkonaut> https://github.com/fokkonaut/F-DDrace/blame/935c834e21e72d3d6154164de8af5ed63709035f/src/engine/server/server.cpp#L2266
09:27 <+bridge> <fokkonaut> https://github.com/fokkonaut/F-DDrace/commit/e56dce3fe6e65d3ec08836acd042b8a34f4b6820
09:27 <+bridge> <fokkonaut> 6 years ago:
09:27 <+bridge> <fokkonaut> https://github.com/fokkonaut/F-DDrace/blame/935c834e21e72d3d6154164de8af5ed63709035f/src/engine/server/server.cpp#L2266
09:27 <+bridge> <fokkonaut> https://github.com/fokkonaut/F-DDrace/commit/e56dce3fe6e65d3ec08836acd042b8a34f4b6820
09:43 <+bridge> <Assa> I checked again, same map, I don't know if ZenBlock is yours, but maybe I misstook afk players for bots, one 'player' is a bot, but is supposed to be a statue of r1 or something, none of the bots are listed, there is a player currently online there with 3 dummies - I need to check my old screenshots again
09:45 <+bridge> <fokkonaut> How do you even come up with all that bs?
09:46 <+bridge> <Assa> I can't remember what I wanted to say back then
09:46 <+bridge> <fokkonaut> > statue of r1 or something
09:46 <+bridge> <fokkonaut> Can you show a screenshot?
09:48 <+bridge> <Assa> this guy, with a ping of 0, all other bots have the same skin
09:48 <+bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1491706473175056415/Screenshot_20260409-094824.png?ex=69d8ab69&is=69d759e9&hm=51ccc3dbc21ef1a6ea801d77008c7ed6d1459bf150e9bbe1bc95ce1b8a4b77e6&
09:49 <+bridge> <fokkonaut> That's just a bartender bot.
09:53 <+bridge> <fokkonaut> > I don't know if ZenBlock is yours
09:53 <+bridge> <fokkonaut> As a maintainer you should probably know by now. You brought this up in the past already. The community is still labelled `fokkonautsplayground`, indicating that it is indeed my server.
09:55 <+bridge> <Assa> I really can't remember what I wanted from you back then, I put that so casually in the conversation there out of context, I believe I checked something like the scorebord on your server and noticed a problem with bots?
09:55 <+bridge> <Assa> 
09:55 <+bridge> <Assa> Sorry for the fuzz, I don't ping players randomly for no reason
09:59 <+bridge> <fokkonaut> You probably had your reasons
10:06 <+bridge> <Assa> @fokko maybe I meant this
10:06 <+bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1491710893287800842/Screenshot_20260409-100612.png?ex=69d8af87&is=69d75e07&hm=57ba6b4fe0345f3c1834bfafcfc4ac3ac75cfcc3722f158ef5b363ae3e41fdeb&
10:08 <+bridge> <fokkonaut> Please describe what I should see here.
10:08 <+bridge> <Assa> here it shows 21/122 players, when there are only 12 humans, but this might honestly be a
10:09 <+bridge> <Assa> ddnet issue
10:09 <+bridge> <fokkonaut> https://github.com/ddnet/ddnet/blob/04f1bc983ac67ffd36b3de2634b10d02b2d0b8b4/src/game/client/components/menus_ingame.cpp#L806
10:12 <+bridge> <Assa> Maybe this is the reason I falsely thought your server is reporting too many players
10:13 <+bridge> <Assa> like the bots don't take any of the 122 slots, do they?
10:14 <+bridge> <fokkonaut> So you counted players in scoreboard, checked ingame serverinfo and noticed it shows too many and you did not compare it to the browser and then pinged me for being offensive?
10:15 <+bridge> <fokkonaut> There are 6 bots, 128-6=122. If you saw my 6 year old linked commit previously
10:15 <+bridge> <Assa> I falsely assumed this was equal to the server browser number
10:16 <+bridge> <fokkonaut> That must be what you were thinking back then
10:17 <+bridge> <Assa> As I said, I can't remember 🙈 it's not my goal to annoy you, even if it may look like it to you
10:18 <+bridge> <fokkonaut> ? I just confirmed your statement about your false assumption
11:15 <+bridge> <bamcane_tee237> You can also got a 64/16 in some RPG servers
20:10 <+bridge> <milkeeycat> @ryozuki how much code was AI generated in https://github.com/edg-l/blitz ?
20:10 <+bridge> <ryozuki> do u spy me
20:10 <+bridge> <milkeeycat> yes
20:10 <+bridge> <milkeeycat> https://cdn.discordapp.com/attachments/293493549758939136/1491862955602743507/image.png?ex=69d93d26&is=69d7eba6&hm=d16d91f186a09afaf3d0edf902fb9dfefe5f39b6ced8263c0d0bbdbb61ed5476&
20:10 <+bridge> <ryozuki> xd
20:10 <+bridge> <ryozuki> no comment
20:11 <+bridge> <ryozuki> @milkeeycat do u use egraphs in ur compiler?
20:11 <+bridge> <ryozuki> i recommend u look up papers
20:11 <+bridge> <ryozuki> rly interesting
20:11 <+bridge> <ryozuki> llvm doesnt use them
20:11 <+bridge> <ryozuki> https://arxiv.org/pdf/2602.16707
20:12 <+bridge> <ryozuki> its a 2026 paper
20:12 <+bridge> <ryozuki> i wanted to experiment
20:12 <+bridge> <ryozuki> > Recent algorithmic advances have made equality saturation an appealing approach to program optimization
20:12 <+bridge> <ryozuki> > because it avoids the phase-ordering problem. Existing work uses external equality saturation libraries,
20:12 <+bridge> <ryozuki> > or custom implementations that are deeply tied to the specific application. However, these works only
20:12 <+bridge> <ryozuki> > apply equality saturation at a single level of abstraction, or discard the discovered equalities when code is
20:12 <+bridge> <ryozuki> > transformed by other compiler passes. We propose an alternative approach that represents an e-graph natively
20:12 <+bridge> <ryozuki> > in the compiler’s intermediate representation, facilitating the application of constructive compiler passes that
20:12 <+bridge> <ryozuki> > maintain the e-graph state throughout the compilation flow. We build on a Python-based MLIR framework,
20:12 <+bridge> <ryozuki> > xDSL, and introduce a new MLIR dialect, eqsat, that represents e-graphs in MLIR code. We show that this
20:12 <+bridge> <ryozuki> > representation expands the scope of equality saturation in the compiler, allowing us to interleave pattern
20:12 <+bridge> <ryozuki> > rewriting with other compiler transformations. The eqsat dialect provides a unified abstraction for compilers
20:12 <+bridge> <ryozuki> > to utilize equality saturation across various levels of intermediate representations concurrently within the
20:12 <+bridge> <ryozuki> > same MLIR flow.
20:12 <+bridge> <ryozuki> abstract
20:12 <+bridge> <milkeeycat> is there an explanation for idiots?
20:12 <+bridge> <ryozuki> ask ai
20:13 <+bridge> <ryozuki> but basically
20:13 <+bridge> <ryozuki> when u do a optimizing compiler with phases
20:13 <+bridge> <ryozuki> like llvm
20:13 <+bridge> <ryozuki> u peep to choose the route kinda
20:13 <+bridge> <ryozuki> the egraph allows u to not choose a route kinda afaik
20:13 <+bridge> <ryozuki> so u always find the best
20:13 <+bridge> <ryozuki> for example
20:13 <+bridge> <milkeeycat> I'll take a look at it
20:13 <+bridge> <ryozuki> in llvm o3 is a list of phases in some order
20:13 <+bridge> <ryozuki> the order is kinda arbitrary, well its studied
20:13 <+bridge> <milkeeycat> iirc I've seen a module named egraph in cranelift
20:14 <+bridge> <ryozuki> but there may be a better order etc
20:15 <+bridge> <ryozuki> > When a compiler optimizes your code, it applies a sequence of transformation passes (e.g., constant folding, dead code elimination, inlining). The problem: the order matters. Pass A might enable Pass B, but Pass B might block Pass A if run first. There's no universally correct order — it's a combinatorial mess. This is called the phase-ordering problem, and it's been open for ~45 years.
20:15 <+bridge> <ryozuki> > One elegant way to sidestep phase ordering is equality saturation. Instead of committing to one transformation at a time, you:
20:15 <+bridge> <ryozuki> > Represent your program as an e-graph (equality graph).
20:15 <+bridge> <ryozuki> > Keep adding all equivalent ways to express each computation — without choosing between them.
20:15 <+bridge> <ryozuki> > Once you've saturated (no new equalities to add), then extract the cheapest version.
20:15 <+bridge> <ryozuki> > The key insight: you never throw away a representation. You accumulate equivalences and pick the winner at the end. This avoids the "applying pass A destroyed the opportunity for pass B" problem entirely.E-graphs are the data structure that makes this possible. An e-graph groups equivalent expressions into "e-classes". If a + 0 == a, both live in the same e-class. Rewriting rules expand the graph; extraction chooses the best expression.The famous
20:15 <+bridge> <ryozuki> ai resume
20:15 <+bridge> <ryozuki> for dummies
20:15 <+bridge> <fokkonaut> ryo the viber
20:16 <+bridge> <fokkonaut> you saw Claude Mythos?
20:16 <+bridge> <ryozuki> ye
20:17 <+bridge> <ryozuki> btw im only talking about the project cuz u mentioned it
20:17 <+bridge> <ryozuki> i prefer to not talk about my stuff here anymore
20:17 <+bridge> <ryozuki> xd
20:17 <+bridge> <ryozuki> it looks like a step towards overlords with money
20:17 <+bridge> <fokkonaut> Yea
20:18 <+bridge> <peterwrightq582> omg join girl in cam discord.gg/hcQ4R7mP6x
20:18 <+bridge> <fokkonaut> But it also shows the true power this has
20:18 <+bridge> <ryozuki> ye
20:18 <+bridge> <ryozuki> would love to try it
20:18 <+bridge> <fokkonaut> Ha💸 Ha💸
20:18 <+bridge> <ryozuki> but maybe its marketing fearmongering
20:19 <+bridge> <fokkonaut> You can make just spent tens of thousands of dollars on hardware and build something locally
20:19 <+bridge> <fokkonaut> You can just spent tens of thousands of dollars on hardware and build something locally
20:19 <+bridge> <fokkonaut> You can just spend tens of thousands of dollars on hardware and build something locally
20:19 <+bridge> <ryozuki> @milkeeycat im also making it to have a compiler in my own os, edos
20:19 <+bridge> <fokkonaut> to some extend, yes
20:20 <+bridge> <ryozuki> ill create holyrust
20:20 <+bridge> <milkeeycat> epyc
20:24 <+bridge> <kebscs> all marketing
20:24 <+bridge> <kebscs> every new model they say its agi and going to rule the world or whatever
20:24 <+bridge> <ryozuki> they arent releasing this one to public at all
20:25 <+bridge> <kebscs> to companies then?
20:25 <+bridge> <kebscs> or full gatekeep
20:26 <+bridge> <ryozuki> https://www.anthropic.com/glasswing
20:27 <+bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1491867116754964621/image.png?ex=69d94106&is=69d7ef86&hm=87afab22cbee55c7a3ee3aecbf50eb38b73c26d81986eaea51b4f0c60f94b198&
20:27 <+bridge> <ryozuki> these i think
20:51 <+bridge> <rekinwielorybik> Please add noto emoji. I have verry few emotes that i can use
20:52 <+bridge> <alw5> tclient also has noto emoji so some people on tclient may use emojis that ddnet users cant see
20:52 <+bridge> <alw5> not really a problem but it can be annoying
21:02 <+bridge> <ryozuki> its like some weird elite company testing the model cuz they fearmonger it can find 0 days easily
21:03 <+bridge> <ryozuki> but i think they also wont release this super model to public use cuz china distills their models
21:03 <+bridge> <ryozuki> they will use it to make smaller models or something maybe
21:03 <+bridge> <ryozuki> idk
21:08 <+bridge> <kebscs> its not like they have some crazy tech
21:08 <+bridge> <kebscs> probably just threw more data at it
21:08 <+bridge> <ryozuki> idk
21:08 <+bridge> <ryozuki> they are the frontier in this
21:08 <+bridge> <kebscs> with all the money they recently got
21:08 <+bridge> <ryozuki> more than openapi
21:08 <+bridge> <ryozuki> by far
21:08 <+bridge> <ryozuki> but i guess its natural they wont make all models public when they get to a point or smth idk
21:09 <+bridge> <ryozuki> but idk whats the point of announcing smth we cant use
21:10 <+bridge> <louis> investor hype
21:19 <+bridge> <kebscs> yea ai's main revenue is investors
22:51 <+bridge> <kebscs> funny bug
22:51 <+bridge> <kebscs> why would it be coded for all layers
22:58 <+bridge> <robyt3> From what I can tell it is already present in https://github.com/ddnet/ddnet/commit/72c06a258940696093f255fb1061beb58e1cdd0b
23:27 <+bridge> <cyberfrighter> i thought that was intended or something
23:28 <+bridge> <kebscs> btw why is this the last commit, cant we go further into teeworlds?
23:30 <+bridge> <kebscs> nvm there are older commits