06:13 <+bridge_> <ryozuki> https://vxtwitter.com/MistralAI/status/1733150512395038967
07:05 <+bridge_> <milkeeycat> Rust does have ASan support https://doc.rust-lang.org/beta/unstable-book/compiler-flags/sanitizer.html#addresssanitizer
07:34 <+bridge_> <teero777> no edges? xd
07:34 <+bridge_> <teero777> https://cdn.discordapp.com/attachments/293493549758939136/1513778319852699750/image.png?ex=6a28f76c&is=6a27a5ec&hm=38cb8c7920627ece88a14cf5c648d165560d97197f9458b9349a6f40077c65f3&
09:02 <+bridge_> <mpft> i dont know what you measn
09:02 <+bridge_> <mpft> i dont know what you mean
09:27 <+bridge_> <tojimolly> what to do if you are blocked on the map
10:08 <+bridge_> <jxsl13> make the t have a flat top, not a pointy one!
10:15 <+bridge_> <chillerdragon> This is art
10:29 <+bridge_> <alexvae1> https://cdn.discordapp.com/attachments/293493549758939136/1513822404332490762/1.jpg?ex=6a29207a&is=6a27cefa&hm=ecee6be072c7343d4b033f25151fd91b2160db1fe88fb091b288801bd21ce1a1&
10:29 <+bridge_> <alexvae1> https://cdn.discordapp.com/attachments/293493549758939136/1513822404764237906/2.jpg?ex=6a29207a&is=6a27cefa&hm=8992b5d59f0aa54b67cea99e5c145d72fe6c3ab02a04c260b34f062d2afc179c&
10:29 <+bridge_> <alexvae1> https://cdn.discordapp.com/attachments/293493549758939136/1513822405129404546/3.jpg?ex=6a29207a&is=6a27cefa&hm=748220fb39f50afd14bbb1d54fde0c35d3a317a150161fcfcef5c469824c6ff8&
10:29 <+bridge_> <alexvae1> https://cdn.discordapp.com/attachments/293493549758939136/1513822405468885093/4.jpg?ex=6a29207a&is=6a27cefa&hm=6ddf4b1d6b47cae31a5cb6c8525ab374fd10bf6cde49a322fb45e34959e3cf42&
11:11 <+bridge_> <Assa> @chillerdragon what's this syntax, never saw this:
11:11 <+bridge_> <Assa> ```C++
11:11 <+bridge_> <Assa> new(pPersistent) CPersistentClientData();
11:11 <+bridge_> <Assa> ```
11:11 <+bridge_> <Assa> so I assume this takes the type of pPeristent
11:12 <+bridge_> <Assa> but shouldn't it be without the brackets at the end?
11:13 <+bridge_> <Assa> this is just a constructor call 🤔
11:14 <+bridge_> <Assa> I don't get the PR
11:18 <+bridge_> <bamcane_tee237> new(pPointer) to use the memory of the pointer points instead of requesting new memory.
11:19 <+bridge_> <bamcane_tee237> @essigautomat  new(pPointer) to use the memory of the pointer points instead of requesting new memory.
11:21 <+bridge_> <timur1234868> nbvt fqgb
11:27 <+bridge_> <Assa> this this reconstruct the pointer then?
11:27 <+bridge_> <bamcane_tee237> yeah
11:27 <+bridge_> <Assa> 🤯
11:27 <+bridge_> <chillerdragon> @essigautomat: it’s called placement new iirc
11:30 <+bridge_> <chillerdragon> I actually have used it before somewhere forget for what tho
11:31 <+bridge_> <chillerdragon> Make sure member init with = in the class works for persistent data if it is not set in the persist method. By calling the constructor for the memory that was allocated with a raw malloc call
11:31 <+bridge_> <Assa> I get the PR now
11:31 <+bridge_> <chillerdragon> I figured since you merged it xd
11:32 <+bridge_> <Assa> this basically moves the constructor logic into the function here and I understand you need a non-default constructor maybe for 7777
11:32 <+bridge_> <Assa> since this was a struct with fields which you just fill with data
11:32 <+bridge_> <chillerdragon> Yes big 7777
11:32 <+bridge_> <chillerdragon> But even default constructor doesn’t work
11:33 <+bridge_> <chillerdragon> It’s random garbage memory
11:33 <+bridge_> <Assa> I wonder if a generic constructor would be possible
11:33 <+bridge_> <chillerdragon> Until it’s set in the persist method
11:33 <+bridge_> <Assa> I guess it would be with an interface an virtualization 🙈 but lets not overkill
11:33 <+bridge_> <chillerdragon> Just make it a virtual init method
11:33 <+bridge_> <Assa> I guess it would be with an interface and virtualization 🙈 but lets not overkill
11:34 <+bridge_> <chillerdragon> Yea I can survive like this
11:34 <+bridge_> <chillerdragon> This pr already saves me some nasty patching
15:36 <+bridge_> <milkeeycat> https://datatracker.ietf.org/doc/html/rfc6648
15:36 <+bridge_> <milkeeycat> > Historically, designers and implementers of
15:36 <+bridge_> <milkeeycat> >    application protocols have often distinguished between standardized
15:36 <+bridge_> <milkeeycat> >    and unstandardized parameters by prefixing the names of
15:36 <+bridge_> <milkeeycat> >    unstandardized parameters with the string "X-" or similar constructs
15:36 <+bridge_> <milkeeycat> >    (e.g., "x."), where the "X" is commonly understood to stand for
15:36 <+bridge_> <milkeeycat> >    "eXperimental" or "eXtension".
15:36 <+bridge_> <milkeeycat> I never thought `X-` meant anything, TIL 😬
15:46 <+bridge_> <s0.0la> never heard "xtreme" or "xtra"!?
16:01 <+bridge_> <teero777> X11 😬
16:26 <+bridge_> <teero777> @patiga btw your mapview is extremely heavy xd https://mapview.patiga.eu/view/?kog=Mixi4Karolina
16:26 <+bridge_> <teero777> https://cdn.discordapp.com/attachments/293493549758939136/1513912119874031646/image.png?ex=6a297408&is=6a282288&hm=bce7d406b1623357da571d5fabdc67b77217e33f41ee6d130a8a53ba5d50c4a0&
16:26 <+bridge_> <teero777> https://cdn.discordapp.com/attachments/293493549758939136/1513912120159506512/image.png?ex=6a297408&is=6a282288&hm=217c194170efefcbd0f1f15ca83f80c1b838f4f76d369348942630b42dd07fe1&
16:26 <+bridge_> <teero777> https://cdn.discordapp.com/attachments/293493549758939136/1513912120587194418/image.png?ex=6a297408&is=6a282288&hm=06ee0d230dd3657baf16cd9f9536c49048b0f61e93c89612e2e085c88f2b1992&
16:27 <+bridge_> <patiga> please elaborate what the screenshots are ^^
16:28 <+bridge_> <Assa> looks fine to me, no insane animations on the map :justatest:
16:28 <+bridge_> <teero777> why does it take 3x more of my cpu than default ddnet xd
16:28 <+bridge_> <teero777> why does it take 3x more of my gpu than default ddnet xd
16:29 <+bridge_> <patiga> ah 1. is twgpu/mapview, 2. is idle and 3. is default ddnet?
16:29 <+bridge_> <teero777> yes
16:29 <+bridge_> <teero777> but ddnet with 800fps, not 170
16:29 <+bridge_> <patiga> could you do a comparison zoomed out?
16:29 <+bridge_> <Assa> might be shader related
16:29 <+bridge_> <teero777> uh let me check what ddnet is with vsync on
16:30 <+bridge_> <teero777> they are both full zoomed out
16:30 <+bridge_> <teero777> the whole map is visible
16:30 <+bridge_> <patiga> interesting, could you do one more screenshot with chromium where the mapview will actually use webgpu?
16:30 <+bridge_> <patiga> (to verify that webgpu is enabled just see if weapons are rendered as pickups)
16:31 <+bridge_> <patiga> (or look into the console in the adapter info)
16:33 <+bridge_> <teero777> uh this is chromium
16:33 <+bridge_> <teero777> https://cdn.discordapp.com/attachments/293493549758939136/1513913886393499770/image.png?ex=6a2975ad&is=6a28242d&hm=80270061f1f0985612900f54488632175a044f3b567d83481441ffab18f4b7ca&
16:34 <+bridge_> <teero777> performance is the same
16:35 <+bridge_> <patiga> no webgpu sadly
16:36 <+bridge_> <patiga> maybe another interesting data point would be the wasm-compiled ddnet client
16:36 <+bridge_> <patiga> which will then also use the same graphics backend
16:39 <+bridge_> <patiga> my rendering implementation only substantially diverges from ddnet for tilemap rendering. on typical zoom levels ddnet should be more performant, and from some zoomed out level twgpu should win
16:39 <+bridge_> <patiga> ah, the native`twgpu-map-inspect` would also be interesting
16:39 <+bridge_> <Assa> do you reuse the same shaders and do you make use of the quad grouping?
16:40 <+bridge_> <Assa> like I can imagine some edge cases
16:40 <+bridge_> <patiga> yes definitely, I diverge there as well, but I wouldn't have expected these changes to be **that** significant performance-wise
16:41 <+bridge_> <patiga> not sure exactly what you mean by shader-reuse, I use the same shader for all quads, and render each quad layer individually with one draw call
16:42 <+bridge_> <Assa> can you render all quads with one draw call even if they're animated?
16:43 <+bridge_> <patiga> yes
16:43 <+bridge_> <patiga> I designed them differently, where actual quad vertices are static and I have an envelope buffer that gets accessed in the vertex shader
16:44 <+bridge_> <patiga> so good call, thats another place where twgpu offloads work from the cpu to the gpu
16:44 <+bridge_> <Assa> I think this might be a better option
16:44 <+bridge_> <Assa> (I mean yours)
16:44 <+bridge_> <patiga> this also means tho that twgpu keeps more buffers so takes up more vram
16:45 <+bridge_> <Assa> the current ddnet client also tries to group quads which has the same envelopes and renders them all with one draw call
16:45 <+bridge_> <patiga> you implemented that system, right? nice one :)
16:46 <+bridge_> <Assa> yes, thanks 🙂
16:46 <+bridge_> <Assa> also we now have clip regions for layers, but if you render everything this should be irrelevant
16:47 <+bridge_> <Assa> but this is one of the reasons the client should be more performant if zoomed in
16:47 <+bridge_> <Assa> because it's specialized for this case
16:49 <+bridge_> <Assa> I also still have <https://github.com/ddnet/ddnet/pull/11303> on my todo list for the zoomed out case ... I am almost finished with reworking the rendering ;_; only a few PRs
16:54 <+bridge_> <patiga> yea the hope of twgpu is kinda that throwing triangles at the screen in this pretty ez way *should* be basically free ^^
16:54 <+bridge_> <Assa> you probably win in any case if you benchmark against ogl1
16:56 <+bridge_> <patiga> eeh, my tilemap solution tends to make me lose pretty much always on normal zoom levels ^^
16:56 <+bridge_> <Assa> why did you not do it the ddnet way?
16:59 <+bridge_> <patiga> I kinda thought about how I'd like to do it, and thats what I came up with
16:59 <+bridge_> <patiga> I think during research I was inspired by some blog post :)
17:00 <+bridge_> <patiga> this one :D https://blog.paavo.me/gpu-tilemap-rendering/
17:00 <+bridge_> <Assa> I am also a bit unhappy how ddnet does this tbh - but this has to do with some iGPUs from intel doing weird stuff
17:01 <+bridge_> <patiga> on my todo-at-some-point for twgpu is properly supporting zooming out at levels where tiles are smaller than one pixel
17:01 <+bridge_> <patiga> tho I should probably do anti-aliasing first
17:02 <+bridge_> <Assa> hmm
17:03 <+bridge_> <Assa> I have yet another optimzation idea, but the implementation is too insane
17:03 <+bridge_> <patiga> I'm all ears ^^
17:04 <+bridge_> <Assa> I mean for the client: you could detect if a tile is a full 1 color tile, and combine tiles to create greater rectangles
17:04 <+bridge_> <patiga> heh, quite specific :D
17:05 <+bridge_> <Assa> yes but often done
17:05 <+bridge_> <patiga> tru
17:05 <+bridge_> <patiga> even in grass_main
17:05 <+bridge_> <Assa> I wonder if I know any map where this is not done
17:05 <+bridge_> <Assa> even my own map 🤔
17:05 <+bridge_> <Assa> since the edges are pasted over full tiles
17:06 <+bridge_> <patiga> ah tru with freeze/teles its pretty much in every rmap
17:06 <+bridge_> <patiga> and probably also other design
17:06 <+bridge_> <Assa> and the detection and combining can be put into the initialization code
17:06 <+bridge_> <Assa> can't wait for the next complaint about map loading times 🙁
17:07 <+bridge_> <patiga> :d
17:07 <+bridge_> <Assa> you can load maps really really fast by disabling lots of rendering optimizations
17:08 <+bridge_> <Assa> I think I might give this a try tbh
17:08 <+bridge_> <Assa> can only be a learning experience
17:09 <+bridge_> <fushi_gg> time to add crazy loading screens
17:09 <+bridge_> <fushi_gg> and optimize even further
17:12 <+bridge_> <ddnet.org/downloads enjoyer> https://github.com/ddnet/ddnet/issues/12264#issuecomment-4661181256
18:09 <+bridge_> <Assa> I actually think this is a non issue, yes we could add the feature but I doubt this would get widespred usage
18:37 <+bridge_> <kebscs> I agree with sans. It's a not high priority issue because it sends a message to the mod aswell
18:37 <+bridge_> <kebscs> But saying it's not actionable is wrong
18:42 <+bridge_> <qetoast> just let the vote happen and ignore the result if authorized honestly
18:55 <+bridge_> <Assa> hmm okay, I guess it is actionable - I just often wish people would propose what they would want instead, which is of course sometimes hard to do
18:56 <+bridge_> <Assa> this would also mark you as a moderator
18:57 <+bridge_> <Assa> but as I said in the general chat, the opinion of the moderators here matters most, as they're the ones using whatever tools we provide them
18:57 <+bridge_> <teero777> the issue is that people can scan for mods but just spamming votes and checking
18:57 <+bridge_> <qetoast> change name on the vote to someone else :kek:
18:57 <+bridge_> <qxdfox> make the vote fail
18:58 <+bridge_> <teero777> fake f4 votes
18:58 <+bridge_> <qxdfox> **
18:58 <+bridge_> <teero777> rigged
18:58 <+bridge_> <Assa> lol just make the player ban himself if he tries to vote a moderator
18:58 <+bridge_> <teero777> please no
18:58 <+bridge_> <melon> :aylmao:
19:16 <+bridge_> <louis> weekly ai news! https://www.anthropic.com/news/claude-fable-5-mythos-5
19:17 <+bridge_> <Assa> @l._ouis monthly reminder about eyes issue :justatest:
19:39 <+bridge_> <Assa> @patiga `AdapterInfo { name: "NVIDIA GeForce GTX 980, or similar", vendor: 4318, device: 0, device_type: Other, driver: "", driver_info: "WebGL 2.0", backend: Gl }`
19:39 <+bridge_> <Assa> 
19:39 <+bridge_> <Assa> I wonder how it arrives at this graphics card, what do you mean _similar_
19:40 <+bridge_> <Assa> also `WEBGL_debug_renderer_info is deprecated in Firefox and will be removed. Please use RENDERER.`
20:02 <+bridge_> <patiga> probably tracking protection
20:02 <+bridge_> <Assa> oO okay I guess
20:02 <+bridge_> <patiga> I got `Intel(R) HD Graphics, or similar`
20:50 <+bridge_> <patiga> rate my text renderer! ✨
20:50 <+bridge_> <patiga> https://cdn.discordapp.com/attachments/293493549758939136/1513978586854457434/image.png?ex=6a29b1ef&is=6a28606f&hm=c2831d68220fb56b39ef8a040645a87701f4737c55a14bcab8cc420653b5b5dd&
20:52 <+bridge_> <milkeeycat> 10 cheese burgers out of 49 twinbops
20:52 <+bridge_> <patiga> I'll take that :D
20:59 <+bridge_> <kebscs> Seems worse to the old model
20:59 <+bridge_> <kebscs> For my use case
21:02 <+bridge_> <ddnet.org/downloads enjoyer> Make the vote actually target the caller
21:02 <+bridge_> <ddnet.org/downloads enjoyer> ngl that would be funny
21:02 <+bridge_> <ddnet.org/downloads enjoyer> I was told you cannot mess up fonts in a way that makes text unreadable, I was wrong.
21:03 <+bridge_> <ddnet.org/downloads enjoyer> I was told you cannot mess up fonts in a way that makes text unreadable, I was wrong to belive that.
21:03 <+bridge_> <ddnet.org/downloads enjoyer> I was told you cannot mess up fonts in a way that makes text unreadable, I was wrong to believe that.
21:04 <+bridge_> <louis> Yes defn. depends on use case, its nice that its not as expensive as mythos was though
21:05 <+bridge_> <louis> im gonna have it bug check the quic implementation i made opus 4.8 do :pepeW:  this is the future
21:08 <+bridge_> <Assa> I love working on rendering, it's always so much fun when stuff goes wrong xD
21:08 <+bridge_> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1513983166661263360/screenshot_2026-06-09_21-07-49.png?ex=6a29b633&is=6a2864b3&hm=a90efbe45c7d159aa853e1f6a635b734627d82ef13bd0b75c8cbf2fbf88781d1&
21:08 <+bridge_> <patiga> :D
21:08 <+bridge_> <patiga> it really is
21:09 <+bridge_> <patiga> but also sometimes disheartening as debug output can be difficult :d
21:09 <+bridge_> <Assa> I am pretty sure my above proposal would work xD I just don't know if I have the willingness to test this
21:09 <+bridge_> <patiga> ooh, you doing the grouping of single-color tiles? ^^
21:09 <+bridge_> <Assa> maybe
21:09 <+bridge_> <Assa> but this would need a kind-of fundamental rewrite
21:10 <+bridge_> <patiga> probably np-hard to get the "perfect" configuration of combined tiles ^^
21:10 <+bridge_> <Assa> I know for a fact that it's np hard
21:10 <+bridge_> <Assa> since I did the pixelart thing ...
21:10 <+bridge_> <patiga> yea 😓
21:10 <+bridge_> <Assa> but greedy should be enough anyway, since we are rendering row by row
21:11 <+bridge_> <patiga> damn :swea != :sweat , as the latter is a perfect match
21:11 <+bridge_> <patiga> wanted 😅
21:12 <+bridge_> <Assa> the issue is, we currently have a one to one relationship between tiles and visuals
21:12 <+bridge_> <patiga> looking more column-by-column up there
21:13 <+bridge_> <Assa> I just skipped every second tile, if the one before is the same
21:13 <+bridge_> <Assa> like as a proof of concept
21:13 <+bridge_> <patiga> heh didnt even notice the broken vertices up there
21:13 <+bridge_> <Assa> the broken vertices are from the skipped tiles still beeing rendered 🙈 with random data
21:13 <+bridge_> <patiga> yup, sounds like graphics programming :D
21:15 <+bridge_> <kebscs> It's 2x tokens and it's still kind of a preview version
21:23 <+bridge_> <robyt3> It's one of the oldest open teeworlds issues: https://github.com/teeworlds/teeworlds/issues/19
21:25 <+bridge_> <ddnet.org/downloads enjoyer> incredible
21:26 <+bridge_> <ddnet.org/downloads enjoyer> Does it qualify as stale or history artifact?
21:28 <+bridge_> <murpi> It burnt through half of me my weekly limit in just 2 prompts :justatest:
21:28 <+bridge_> <murpi> It burnt through half of my weekly limit in just 2 prompts :justatest:
21:43 <+bridge_> <robyt3> There probably would have been more impactful optimizations that could be applied first before attempting this, like those in the DDNet client. At this point the entire repository is stale with the last release being 6 years ago and the last PR merged by oy almost 2 years.
22:25 <+bridge_> <louis> :justatest: i guess i could run tasks for you if interested xd
22:26 <+bridge_> <murpi> @blaiszephyr needed it, not me 😄
22:27 <+bridge_> <melon> :aaa:
22:27 <+bridge_> <murpi> Its likely a bug. They always seem to have some issue that burns through people's tokens whenever they release a new model xD
22:28 <+bridge_> <melon> I once tried one of those in-editor AI things.
22:28 <+bridge_> <melon> 
22:28 <+bridge_> <melon> It repeated the sentence "error opening file" within a minute or two until I hit my 500K token limit for the month :justatest:
22:29 <+bridge_> <melon> I once tried one of those in-editor AI things.
22:29 <+bridge_> <melon> 
22:29 <+bridge_> <melon> It repeated the sentence "error opening file" until I hit my 500K token limit for the month :justatest:
22:29 <+bridge_> <murpi> coding agent moment
22:30 <+bridge_> <melon> is there a reliable one even?
22:30 <+bridge_> <melon> I don't want in-line completions and such, I just like to use AI as a Google replacement sometimes xd
22:30 <+bridge_> <melon> (Claude does a pretty decent job at that)
22:33 <+bridge_> <murpi> automated linting, unit testing, and building :greenthing:
22:33 <+bridge_> <reitw> https://www.youtube.com/watch?v=6YPqoARpYuQ
22:34 <+bridge_> <melon> 🫠 maybe I'm weird but I really hate using AI generated code
22:34 <+bridge_> <melon> 
22:34 <+bridge_> <melon> If I don't fully understand what (my) code does it makes me anxious
22:37 <+bridge_> <kebscs> Tell it to explain it :p
22:39 <+bridge_> <murpi> AGI is here
22:39 <+bridge_> <murpi> https://cdn.discordapp.com/attachments/293493549758939136/1514006114759344260/image.png?ex=6a29cb92&is=6a287a12&hm=137b056cf552e261fe1333766d6886face2265d7bc5eb357e6258a2d2fd4679d&
22:40 <+bridge_> <murpi> https://www.reddit.com/r/ClaudeAI/comments/1u1bmim/comment/oqp51qd/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
22:41 <+bridge_> <murpi> https://preview.redd.it/agi-is-here-fable-5-suggest-me-to-drive-to-the-car-wash-v0-9yt2bdep3b6h1.png?width=640&crop=smart&auto=webp&s=95776b34f4161df63a6c9b35cebe2e566b9cedab
22:42 <+bridge_> <melon> :kek:
22:42 <+bridge_> <melon> https://cdn.discordapp.com/attachments/293493549758939136/1514006672920805517/Screenshot_20260609-224142.png?ex=6a29cc17&is=6a287a97&hm=b81afe2f8330ca73a2b8e7a3c33fb43b25888cbc239c1207e176c5c51c099fd7&
22:47 <+bridge_> <aegisub> this makes you sane
22:47 <+bridge_> <aegisub> watch out, the singularity is coming
22:49 <+bridge_> <Assa> I am not sure this is doable with the vulkan backend in a way where you gain performance 🤔
22:50 <+bridge_> <robyt3> I don't think it's a useful optimization for modern backends
22:51 <+bridge_> <Assa> Combining vertices should still be useful, but I think not the color part
23:40 <+bridge_> <totar> really incredible work its doing
23:40 <+bridge_> <totar> https://cdn.discordapp.com/attachments/293493549758939136/1514021499554500779/image.png?ex=6a29d9e6&is=6a288866&hm=e6a3ff2c278b500fbb52b3d8cbd65a7d3e14e99e748760e161a3046dd3a8f9d3&
23:41 <+bridge_> <totar> :galaxybrain:
23:41 <+bridge_> <totar> https://cdn.discordapp.com/attachments/293493549758939136/1514021745697226984/image.png?ex=6a29da21&is=6a2888a1&hm=f3052d51191461958382ae5d3e4fc5e9ff482b9a8ad660eed6e476d3e83389f4&
23:43 <+bridge_> <totar> why did they post this lol
23:53 <+bridge_> <dumpedsoul> https://tenor.com/view/context-context-hat-i-require-context-soyjak-wojak-gif-11496174374422496290
23:53 <+bridge_> <dumpedsoul> ive only played a bit of factorio so far so i cant tell the probably funny fail here