Sabitlenmiş Tweet
Slava Egorov
27.4K posts
Slava Egorov
@mraleph
👋compiling compilers @Google // overall tech lead for the @dart_lang programming language @[email protected]
Aarhus Katılım Temmuz 2009
605 Takip Edilen11.6K Takipçiler
@mraleph should the performance tracking work for `flutter analyze` too (since it opens a DAS)? Or is it intended behaviour for all calls to show as unknown when I profile it?
English
Slava Egorov retweetledi
We're hiring a manager for the Dart team in Aarhus, Denmark.
Interested folks can apply here: google.com/about/careers/…
English
@EmeraldApps_io @luke_pighetti Nah. It mostly overengineering and lack of taste and ability to keep things both scalable and simple.
English
The insane bullshit in question is keeping multi billion dollar companies afloat. The reason its often being applied to small projects is potential scalability, so just because you dont understand something and are not willing to put in the time to grasp it and instead choose to just use whatever AI spits out, it doesnt mean its bad…
English
coded up a flutter project with claude the other day and its code is gorgeous. none of the insane bullshit flutter devs have been circle jerking around for years. i’d take claude over any junior or mid level flutter dev any day
Jeffrey Way@jeffrey_way
If I’ve learned anything about programming over the years, it’s to be suspicious of any person who repeatedly evangelizes increasingly complex approaches and techniques.
English
@m_feinstein @mahersafadii Because it's two different VMs.
Or turn this around: if it was the same level of interop with two VMs then why even bother compiling to one VM (DEX)?
English
@mraleph @mahersafadii Why it won't get the same level of interop? Would it be an actual limitation, or is it just a performance issue? Like direct memory access VS boxing values into representations?
English
what are the actual downsides of Dart compiling to the os (jvm/LLVM) instead of directly to machine code, like we saw all the advantages that Kmp has because of this, dead easy native interop, are there any major downsides to doing it this way?
cc @mraleph
English
@m_feinstein @mahersafadii Because debug mode will have to use the same target as release - DEX bytecode, otherwise you dont get the same level of interop.
(It's fair to say that you don't truly lose hotreload but it becomes very hard to implement because you don't have as much control over runtime)
English
@mraleph @mahersafadii Why would we lose hot reload? Hot reload could still exist in a "debug" compilation, with bad performance and all other bad things... And for a "release" compilation we get the best ever possible binary (whatever that means)... I only suppose that it increases complexity
English
@ulusoyapps @mahersafadii Poorly? Compose hot reload uses a fork of OpenJDK and does not work on Android.
With enough effort you can make hot reload work on any platform, but the question is: why build house of cards when you already have a pretty good one made out of bricks.
English
English
@mahersafadii If it was a no-brainer we would have done it. It's unfortunately is not no-brainer.
English
@mraleph I used to follow the discussions that you used to have with jim, you did mention that it comes at a performance cost, but how big is it actually, because the trade off being dead easy and low cost interop sounds good
English
@mahersafadii we have built compiler to Java bytecode several years ago - and IIRC even had a simple Flutter app running on it, but in the end tradeoffs are so complicated and unclear that adding yet another target just did not feel right at that point. so we never shipped it.
English
@mahersafadii the simplest example is: you want to do hot reload, now you need to figure out how to do that hot reload on ART. (good luck)
English
@bunopus все, что нужно знать о Уелче, описано в книге The Man Who Broke Capitalism
@badlogicgames I asked of my dear friend orator Prig:
'What's the first part of oratory?' He said: 'A great wig.'
en.wikisource.org/wiki/Page:Life…
English
that's not controversial. that's just dumb.
Rohan Paul@rohanpaul_ai
Chamath on how AI agents are making the "10x engineer" distinction disappear because the most efficient "code paths" are now obvious to everyone. Just as AI solved chess and removed the mystery of the best move, AI is doing the same for coding, making the process reductive and removing technical differentiation. "I'm going to say something controversial: I don't think developers anymore have good judgment. Developers get to the answer, or they don't get to the answer, and that's what agents have done. The 10x engineer used to have better judgment than the 1x engineer, but by making everybody a 10x engineer, you're taking judgment away. You're taking code paths that are now obvious and making them available to everybody. It's effectively like what happened in chess: an AI created a solver so everybody understood the most efficient path in every single spot to do the most EV-positive (expected value positive) thing. Coding is very similar in that way; you can reduce it and view it very reductively, so there is no differentiation in code." --- From @theallinpod YT channel (link in comment)
English
@fmorschel_dev if you have any place where you can upload it - we could try to figure out ways to slice/analyze them in batch mode rather then to try viewing them in UI.
English
@mraleph, I have a big private project I would like to make DAS startup faster. The problem with using the perf command is that the files get huge (~18 min = ~1,5GB) and viewing them is hard. Any ideas on what could we do?
English
you ain't fooling me mr. I-use-slop-generator-to-make-what-I-think-are-useful-and-thoughtful-comments
English
Bearish on pi as a project if the maintainer doesn’t see a path to using his own coding agent to triage and fix issues — what else would pi be *for*?
Mario Zechner@badlogicgames
yeah, i think that's it for OSS. opened the tracker last night. +20 issues in less than 24h. there's absolutely not chance i can make this suistainable, especially since this is a hobby. open to ideas. i can deal with 5-10 issues per day. this is just too much.
English
@HarrySild well, this is at least the type of work I would like LLM do, so there is some hope, so I will keep on trying.
Though from the look of it - still takes longer than doing it by hand and you get questionable outcome, so might still be a waste of time rather than an acceleration.
English
@mraleph Funnily enough Elixir author had the same vibe post not long ago and the community superstars managed to get better results. Not saying skill issue, but there definitely is more to getting stuff out of AI than just asking for the hard problems. x.com/i/status/20198…
José Valim@josevalim
@chris_mccord @pmarreck @pmarreck How the heck did you get Claude to find bugs twice, on Chris' commit and mine, while mine just spinned in circles?! Was it running a debugger? Did it just look angrily at the code?
English
I really wish I was in living in the same parallel universe as AI slop wankers and could just spawn a freaking agent per item of my backlog... And not have to rm -rf the produced garbage later.
English
@kevmoo Some datacenter somewhere burned all those watts and all you got out is a fairly trivial mechanical refactoring - very far away from what I would describe as a true slog.
I am sorry to be a bearer of bad news.
English
github.com/dart-lang/sdk/… this is one of my favorite examples. At the start doing this refactoring didn't make a lot of sense. Then eventually we had a bunch of boilerplate and copied code. You could have done this refactoring by hand but it would have been a slog.
English
Slava, I think you're neglecting all of the toil that most engineers live in everyday. You work at the most complex levels of a complex system. I am not surprised you find very little benefit in the AI generated bits. (Although I am curious about any experiments you've done.)
Slava Egorov@mraleph
I really wish I was in living in the same parallel universe as AI slop wankers and could just spawn a freaking agent per item of my backlog... And not have to rm -rf the produced garbage later.
English