quail
@quail.wtf
A cursed quail. I think too much about databases.
nix-direnv registers gc roots, so by default the things you downloaded for old side projects doesn't get cleaned up
November 30, 2025 at 4:44 PM
nix-direnv registers gc roots, so by default the things you downloaded for old side projects doesn't get cleaned up
Growing up there def gives you a biased perspective, but if you ignore a very healthy amount of political/cultural issues, it's a fun place to spend 2-3 days.
November 23, 2025 at 12:47 PM
Growing up there def gives you a biased perspective, but if you ignore a very healthy amount of political/cultural issues, it's a fun place to spend 2-3 days.
Ok but did she enjoy Missouri, also STL or KC (or the secret third option: Branson)
November 23, 2025 at 12:47 PM
Ok but did she enjoy Missouri, also STL or KC (or the secret third option: Branson)
> I like to think of code review as a tradeoff between shared & solo time, a sync point to distribute & repair context.
Yeah I think what I'm wanting would actually increase this property - the PR itself becomes much more collaborative (between humans, not humans and AI)
Yeah I think what I'm wanting would actually increase this property - the PR itself becomes much more collaborative (between humans, not humans and AI)
November 22, 2025 at 3:31 AM
> I like to think of code review as a tradeoff between shared & solo time, a sync point to distribute & repair context.
Yeah I think what I'm wanting would actually increase this property - the PR itself becomes much more collaborative (between humans, not humans and AI)
Yeah I think what I'm wanting would actually increase this property - the PR itself becomes much more collaborative (between humans, not humans and AI)
Alas I am too young to have lived through a time before code review - but strong agree, it would be amazing if there was _less_ friction to getting stuff into trunk
November 22, 2025 at 3:27 AM
Alas I am too young to have lived through a time before code review - but strong agree, it would be amazing if there was _less_ friction to getting stuff into trunk
Like, yes we should absolutely be having "stuff that AI can catch" be handled in dev - but what does software engineering look like if the AI is doing the coding anyway
November 22, 2025 at 3:24 AM
Like, yes we should absolutely be having "stuff that AI can catch" be handled in dev - but what does software engineering look like if the AI is doing the coding anyway
Yeah to be clear I don't think having AI doing the code review is The Way - I can't imagine a world where we want that.
I _do_ think that AI can change the nature of what code review is and how we go about it
I _do_ think that AI can change the nature of what code review is and how we go about it
November 22, 2025 at 3:24 AM
Yeah to be clear I don't think having AI doing the code review is The Way - I can't imagine a world where we want that.
I _do_ think that AI can change the nature of what code review is and how we go about it
I _do_ think that AI can change the nature of what code review is and how we go about it
(these tweets read like I am an AI maximalist but I'm pretty middle of the road, and tbh if we end up with better code review tools from this bubble that would be nice)
November 22, 2025 at 1:43 AM
(these tweets read like I am an AI maximalist but I'm pretty middle of the road, and tbh if we end up with better code review tools from this bubble that would be nice)
I'm actually really excited about the idea of a "mob review" style - lots of people pointing things out, having a coding agent take the suggestions and implement changes as they come in.
November 22, 2025 at 1:42 AM
I'm actually really excited about the idea of a "mob review" style - lots of people pointing things out, having a coding agent take the suggestions and implement changes as they come in.
I've really only ever used GitHub/gitlab but the whole diff model is just... Not good and generally supposes that the patch is either good to go immediately, or basically needs to be redone. The idea of an iterative refinement is completely foreign
November 22, 2025 at 1:42 AM
I've really only ever used GitHub/gitlab but the whole diff model is just... Not good and generally supposes that the patch is either good to go immediately, or basically needs to be redone. The idea of an iterative refinement is completely foreign
My take is that this is somewhat correct - if we're going to be generating our code more, then we should be putting more investment into making the review step much more robust. IDEs are tools that developers spend lots of time configuring to their needs, what if we did something similar for review?
November 22, 2025 at 1:42 AM
My take is that this is somewhat correct - if we're going to be generating our code more, then we should be putting more investment into making the review step much more robust. IDEs are tools that developers spend lots of time configuring to their needs, what if we did something similar for review?
The year is 365.25 days long, that means we can fall back ~6 times a year.
November 2, 2025 at 6:05 PM
The year is 365.25 days long, that means we can fall back ~6 times a year.
YESSSSS DO IT IT'S SO UNDERRATED AND MAKING YOUR OWN CLOTHES IS EASIER THAN YOU THINK
October 21, 2025 at 11:23 PM
YESSSSS DO IT IT'S SO UNDERRATED AND MAKING YOUR OWN CLOTHES IS EASIER THAN YOU THINK
Ah you can actually see if you're having icache or dcache misses with perf! But also I haven't ran into icache issues yet, so I wouldn't have any advice there 😅
September 3, 2025 at 12:08 PM
Ah you can actually see if you're having icache or dcache misses with perf! But also I haven't ran into icache issues yet, so I wouldn't have any advice there 😅
Back when I was fighting this problem, I just did a bunch of shit to try and use less memory in the hot loop, and then suddenly, as if by magic, I stopped having cache misses on a certain mov instruction.
September 3, 2025 at 12:06 PM
Back when I was fighting this problem, I just did a bunch of shit to try and use less memory in the hot loop, and then suddenly, as if by magic, I stopped having cache misses on a certain mov instruction.
perf can tell you which instructions tend to have misses, but you're almost definitely profiling your release binary & so there's ~zero connection between that and your source 😔.
September 3, 2025 at 12:06 PM
perf can tell you which instructions tend to have misses, but you're almost definitely profiling your release binary & so there's ~zero connection between that and your source 😔.
Tomorrow's headline: Anthropic acquires LD out of spite
September 3, 2025 at 12:02 PM
Tomorrow's headline: Anthropic acquires LD out of spite
It def feels more art than science to me. The main thing is:
- understand how cache lines work
- understand how much memory you're touching during the hot loop
- understand how cache lines work
- understand how much memory you're touching during the hot loop
September 3, 2025 at 11:57 AM
It def feels more art than science to me. The main thing is:
- understand how cache lines work
- understand how much memory you're touching during the hot loop
- understand how cache lines work
- understand how much memory you're touching during the hot loop
shout-out to vulkan-zig for providing a darn good vulkan SDK in a language that isn't C.
And for making their examples the "hello triangle" bit itself so it's uhhhh easy to copy code when I get tired of hand-copying :)
And for making their examples the "hello triangle" bit itself so it's uhhhh easy to copy code when I get tired of hand-copying :)
August 31, 2025 at 3:05 AM
shout-out to vulkan-zig for providing a darn good vulkan SDK in a language that isn't C.
And for making their examples the "hello triangle" bit itself so it's uhhhh easy to copy code when I get tired of hand-copying :)
And for making their examples the "hello triangle" bit itself so it's uhhhh easy to copy code when I get tired of hand-copying :)
Zig is very new contributor friendly 🙏
August 20, 2025 at 10:58 PM
Zig is very new contributor friendly 🙏