Robbie Speed

@VictorTaelin This poll is only going to show the popularity of the models listed, because they all do this shit.

This is a fantastic and deep observation. You have correctly identified a fundamental knot in dependently typed language design. Guess the model!

@JLarky It's worse when you realise the dollar value of hardware is outpacing the performance gains to the point where used 5 year old hardware can be sold for more than it's original purchase price. And to buy equivalent hardware new cost at least double.

"why does software keep getting slower as hardware impoves"

@youyuxi Similar to AI summary comments in PRs and issues. I don't understand why people feel compelled to post them.

The obvious AI slop responses that just rehashes the original post are so insufferable 😩

@munawwarfiroz No worries discussing poor choices in React land is a passion of mine too.

@RobbieSpeed If you ever see me bashing next.js, know that it's not for engagement, it's my passion 😛

Replies, questions on X get zero replies these days. What's the value of this platform? Broadcast messaging? Which too gets random engagement?

@munawwarfiroz In the end even the accounts still on my feed occasionally engage in baiting. I can't really blame them since it's what the platform promotes.

@munawwarfiroz Engagement bait and dog piling are this platforms only successful medium. I've tried conditioning For You feed with liberal use of "Not interested", bait still rises to the top. I switched to Following feed with liberal muting.

@JLarky Thanks ChatGPT I'll stick with Pulumi or Terraform :)

@RobbieSpeed void.cloud

chatgpt is surprisingly insightful today :D

@Jerroudi_Yass @cmuratori @digitalfoundry It wasn't making very good use of whatever data it did have. Like most gen AI video it has jello like contours, and inconsistent lighting. Future versions will probably be better, but so far it's a visual regression compared to previous DLSS versions.

@cmuratori @digitalfoundry What I didn't like about that video, it was short, and it wasn't obvious if DLLS 5 was a filter with no knowledge about the lighting or textures, or did it use data from the game engine.

Just wanted to mention: I watch @digitalfoundry regularly. I saw their DLSS 5 video when they posted it. It was actually the first time I saw DLSS 5. While I was surprised they liked it - since in my opinion it did not look good - that's one of the reasons I watch other people's shows. If I only wanted to hear my own opinions of things, I'd just watch my own channel :) I know it's stressful when a large percentage of your audience is mad about something. I can absolutely understand why Digital Foundry wanted to backpedal a bit on their initial strongly-favorable coverage. But for that same reason, I felt like I should take this opportunity to say: I watch Digital Foundry regularly. I was happy to see their initial coverage of DLSS 5. I had no problem with it, and I couldn't care less if their opinion about it differs from mine. If anything, that's a plus. And furthermore, it seems like both @dark1x and @Dachsjaeger weren't as enthusiastic about it. That actually makes me look forward to more coverage on the channel, since having both people who like and who dislike DLSS 5 should make for some great videos!

@JLarky I'm looking forward to being past the dumb take behind such benchmarks.

Marko keeps winning every benchmark without even trying 😂

@joodalooped @RyanCarniato @solid_js @reactjs LLM smell are patterns that were common pre LLM, and thus common in training data. Reading LLM content has made us hyper sensitive to the patterns. There is very little in terms of writing style that doesn't now trigger the sounds like AI alarm.

@RyanCarniato @solid_js @reactjs "they sound like the most logical person I've come across in response to this" funny consequence of them repeating back the thing they've just read lol i believe you, there were just moments in there like "no X, no Y, no Z" that had LLM smell

People won't like it, but some truths are undeniable. "Two React Design Choices Developers Don’t Like—But Can’t Avoid" by @RyanCarniato #DEVCommunity @solid_js @reactjs dev.to/playfulprogram…

@artalar @devagrawal09 @RyanCarniato I'm not sure I follow. Are you saying on the n+1 run of derivation you should not unsubscribe from the nth run deps? Metron supports async deps, and I've seen no issue with unsubscribing from old deps the moment a derivation is marked dirty.

@RobbieSpeed @devagrawal09 @RyanCarniato The problem with async tracking is that you need NOT to unsubscribe from async deps (after await) while a new rerun is pending

Note, the price of "suspense" is idempotence. It is extremely important - If you are trying to redefine the default language semantics, you will need to recreate all language conditions for this new semantics. With "suspense," you simplify defaults, but harden any branch code.

@devagrawal09 @artalar @RyanCarniato It may not be that simple. I could see it being extra tricky to get async tracking working with Solid 2.0 entanglement.

@artalar @RyanCarniato dependency tracking stops working after the first `await` since there's no async context in the browser yet once async context arrives you should be able to do this

@brenelz @convex If convex exposed an async iterator for subscriptions it could be much cleaner.

What if subscribing to @convex was this easy?!

@JLarky One interesting thing about some of the large models is how much information they contain that you can't even find search results for on the web. However if you're not a expert on the subject it's impossible to tell if it was made up.

We have had access to LLMs for 7 years now. It's crazy that we didn't get any sort of discoveries of anything new in that time It just gets more and more efficient at searching the average result It's like that yellow tint of gpt images, but in knowledge work

@adamdotdev @bleuonbase Thankfully Chromium web view is in the works for Tauri. #issuecomment-3267172258" target="_blank" rel="nofollow noopener">github.com/tauri-apps/cef…

@bleuonbase Because WebKit is absolutely awful

why would they choose electron over tauri in 2026

@filpizlo The quality of types varies a lot across and within ecosystems. Nothing worse than dealing with a library that has incorrect types that conflict with docs and implementation. I imagine this would lead to some unrecoverable scenarios for many coding agents.

@filpizlo It's safe to say provided the types are correct, they do aid productivity when *consuming* typed code/libraries. Whether that's enough to outweigh the burden of *producing* your own types is a case by case situation.

One of the most broken memes in programming languages is that types aid productivity. They don't. Types are a preference, at best

@DelaneyGillilan @RyanCarniato @benapatton You think they would be better as a native app, but it's that kind of functionality that gravitates people toward SPA. If you think the web sucks for this use case you are right, but also wrong. The distribution benefits vastly outweigh the downsides (and there are many).

@RyanCarniato @benapatton When I see what most people build with SPA paradigm they'd be much more well served with either hypermedia or a native app.

Glad to see other people are staying to do independent tests on it. We won every single category and that's even in PHP! How is this not the norm for web dev? zweiundeins.gmbh/en/methodology…

@josevalim @KenAKAFrosty Something really bothered me with this solution which is it would break if a tuple was provided to take, but the tuple had a union in it, so I fixed it. #code/C4TwDgpgBAwg9gOwMYCcLAgHgAoBopQCqUAvFNgHylFQQAeGCAJgM7kBQBBA-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-JY5h1hu6zzAIcI7RQAbHmNx4B8IvCMg63VXxemmYyU8ZghdEYoevn+OIoyojEsRxSIEHXAkiRJfFoBYBBAjASBgDYKZUk7el2AAhwyTYLwmGg-EsXzDtvGyelGWpWl63ocIxjAVIIAARkyZtIylBMRhYTREzqJhlHo-AEGoiBlDHABBVJAiQCAx3wCAxnCQJUmE8JxMkgABKiaLogA6LwxmkkViWUABmAAGE90ygOiE1jCYOOAWYCHLTIAFlwlJLSUEiFIxiYqBMCgEytIAVmY2QbLs6g2VyTJ2M0fA+DHASxik-Ax1k+TUjHQRihfWhUg40suHC9BIvZGKOJQeLEsE-Sx3CYkspyghTxkD8vygYr7MGdgcwTBjqA02iGKYwwGK0pKIBRNFMPJHE4Fwjr8LjJgoDa+tSWgwTVogWN4xTEUJjgDRSKpZlKLoaihoAJl8lteohCquMCHioD4lRBOEsSJJS2g5IUpSVIgdSLs0iAdLgPSIkMqBTPMmQrIaP0HLqfIXLc-EPK8iHfP8wKQsKghGjYMgEom2r0oUrL9t63KTQKltCb9ahSZq1L6qk4RwmwqNgBfIx33QT91rK2LKsRuwj1Pe6ICugaQeukb2Bl8bBKmqAADFlNSCZwiQbJWTgKAwHCFAaWUk6mQokRLroozbpY6VRael63omz7Af0inFKgOrAeBm2wd0-T2eMsykXhkqieRpzXPczzmGxktceC0KuCJ5nquS8m-syznudY2n8ugBnxeJ-hM5+uqGrzuaC+a-nWsF9qoogcq4tLrS5LATIchLEgqGyJs-2loy5YDu2jwgIyVeS-QoAAiajbgFgWECCZlUhXCYVUJh-zRAAJLy6NZEYhqgDtAlJCFYMQBlTqtwa6IAFntgh7qdqpuN4-iPp9r7JM9jnAG31-ag3BpDAyQkYZhwsgjRo0dUaxwxvHbyOMAopwJuLagaUc6ri5rXfwFlkx02LseRo2D2Z4PzoQvmUBG7ACFqyEWFUqpk3wI0bKw8ea0CfmPUGL9J5PxnpNIAA" target="_blank" rel="nofollow noopener">typescriptlang.org/play/?#code/C4…

@josevalim @KenAKAFrosty You can fallback to a partial in that case. That would be the correct thing to do when receiving generic array types rather than tuples, even a type like `"a"[]` might not actually contain an item with "a", because the array can still be empty. #code/PTAEE0HsFdQYwIYDt4AtkHMCmoAuqBLAZ1ADNok5cDIkAuAKHMutrwQGssAeB0UACqgsAD1xYkAExIAlLHEgAnSdyK5FBJBgA0oZAE8AfNr7xaa0AGlhYidNCKsCSbQA2+0AAou+yKUEAlADaALoMhp6mkABGAFZ0gib8pARYrtIJlgwBmTbiUiQGoAD8oEjQALbRWIp5diSWQQBErhIY+E0hJaAACgiK1Aiu3D0EcBzcArqN5VU1IYaGoAmj45PTQbPVigvLZVgAbjWgAN6mCkgWjkTQrrigALynAL56hUj6ANympqRKXhcLCk0pJQH4yKl0kQAqdTPwCP5PABCTzA9KgTRguIBGFnfj4vCoRSQADu+zJAFFFMTFJ4AAaWLAeJoAEhOAGV1JoMKjIZIAs8mmVIPc-hRQZiYrF5Lg6QFvgTnnCHFgbncgmjJF0nlKNXyQgrQEr+KZHLhoIoUNdbrhvkqGCAIDB4MgkCK0JgcPgcEQkAQAA7+rC4EjVVykxgOsACQgkBSSH2oGDoklKDiMZhUGgoUQICr+1oARk8uPO5nu0CIxyeJwxkgShd0SDzWASTQAgq4xlgmrosBUEARXG2ht2AAK5-OtAB0Cgqvb02ASAGYAAxGw2te43aJV20MUwIzwAWQQ+GnimQLgqJdA3FAq+nAFZSwSd3vHuwuJ5KzVdM1mwqHtdCaftB1cToAlMV40irWE32gXdg0-XBOCwH8q0Uf8mkA4DQCaBBsEg6CD34M0LRQd9gztA9HRjYgzATUAiCTW5QXIy1CR9FsVTVe4EEKaJICODMKCzNhJwLLAACZbzxMxLgrTDP1rAh61ARsyhbNtO27BcwKHEcuzgLAJxEPMpNnSB510QjW1ANcN1MLdQB8IhvkPUgTzPVALyvazb3vR8X3g-E3M-ACWwXUCByHTo3gUtRoOEVw4Pk-hwqeSKgOiuz4oExLcBIvhTWDCiv3Q38sNcploRoqNQAAMSGVxogQcY8EgUB-X6QZhyYMTWBzcypywZc5LLRTQCqlS6wbJttPw3STP02Lh3w0cTLMiyZznBc7JXddnk3ZC3I8+EvNPc9LykAKYSC59XzC2qIpwqKQIMiCugKwEiv4GDUpwdKav0EgsrenKQLy7643LYrQFK81ONQ78qt0NzpwHf1PG8GEHiWDgcRooA" target="_blank" rel="nofollow noopener">typescriptlang.org/play/?#code/PT…

Most people treat the static vs dynamic debate as a foregone conclusion, but in practice type systems restrict the expressive power of programming languages in ways that leak complexity to developers. New article: Type systems are leaky abstractions — the case of Map.take!/2