Notices by Hailey (hailey@hails.org), page 3

@whitequark @amy its fine and even good imo to have a slow and rigorous CI that builds from scratch and can catch this stuff, just don’t run it in band of the dev loop

@xssfox @whitequark the best and fastest CI I ever used was jenkins that it was someone else’s job to manage. On the other hand I tried to set up jenkins myself a little while ago and it was absolutely inscrutable

@whitequark @tef and you gotta manually specify dependencies - another potential source of errors - because the reactivity doesn’t extend so far that it can figure out the dependencies for you

@whitequark oooh what are you building?

@tef @whitequark and a second whole class of bugs where you can really easily introduce accidental re-renders from all the footguns in the hook system

for example <MyComponent onEvent={() => …} /> will trigger a recursive re-render every single time because a new function instance is created right there. you gotta pull the closure out, wrap it with useCallback, and pass the memoised callback to onEvent instead.

and with all that boilerplate to do things correctly, what are even we gaining?

@whitequark pages and pages of spec and no clear benefit over gzip

Do you use git gui or other Tcl/Tk apps and find yourself annoyed that your staid old unix program does not look very nice on your beautiful new hidpi monitor?

Simply LD_PRELOAD=tk-hidpi.so and enjoy Tcl/Tk in gorgeous high definition! https://codeberg.org/hails/tk-hidpi

I really like using a magic trackpad on my desktop but I've gone through so many of them over the years where the haptics eventually stop working. I just always assumed it was a Linux problem though I never really came to a satisfying explanation of the problem.

Anyway apparently this is a known issue caused by the battery swelling up when you leave them plugged in permanently 🙃 (my trackpads also stopped sitting flat on the desk but I never considered that this might be related) https://discussions.apple.com/thread/255223984

maybe i'm too much of a boomer or too much of a digital native or whatever but I prefer wired connections for peripherals and it seems you're just not supposed to do it that way? My experience with wireless shit is that there's always annoying latency spikes and general unreliability, and I am unfussed by wires otherwise

btw it’s not just the lead hyprland dev who’s toxic, it’s the whole community. Go check out the official project discord and you’ll see it’s full of pepe frogs and the vibe of 4chan boys posting the most repugnant shit you’d expect them to, all the way back to the very beginning. That discord is functionally a radicalisation funnel. If you’ve been around the internet long enough you know exactly the vibe as soon as you see it.

+ btw this is how they’re talking in DHH’s official Omarchy discord.

They fully know what it’s all about. There’s no ambiguity here.

@whitequark @wingo like most of these technologies out of big tech its a tool for deskilling labour and increasing labour arbitrage. just imo!

Was checking some stuff in the local laws and discovered it's not just allowed to create a nuisance with a bike, it's mandatory

when trusted with autoplay, discord for some reason opens an audio playback stream just to play silence which causes pipewire to sit at 2% cpu with its alsa timer wakeup thread making about 750 syscalls per second, even when the tab is not visible

tip: if you use discord.com in firefox on linux, disable media autoplay to save about 1w of continuous power draw

I'm not really a web maximalist, but I'm a firm believer that web apps can be fast. Peek under the hood of any mature UI toolkit and you'll see it's all the same stuff. They all have node trees, dynamic layouts, event bubbling, etc. GTK even has actual CSS! The problem with the web is cultural, not technical. It is just overwhelmingly common for web apps to do really daft things

Had a support ticket open with Vultr requesting a refund for a couple month's worth of charges for a cloud storage bucket I couldn't delete due to errors on their end. I said I'd begin the chargeback process with my bank if they couldn't refund me and they nuked my account on the spot! Glad there wasn't anything important left in there

the problem with elastic cloud services is that they will scale as far as your wallet does. But usage running away outside of predicted growth is always something I want to know about. normal servers hit a wall sooner (out of disk, out of memory, etc) and raise your attention so you can go and fix them. cloud shit just runs up your bill and you find out well after the fact

every time I've deployed something """properly""" with cloud shit it has ended up costing a bomb while having mediocre performance and reliability. conversely every time I've deployed something to just a normal debian server it has been rock solid with no issues and no ops effort required for years

the cloud is just Worse than well-run normal linux servers. but running your business on well-run normal linux servers yields over too much power to labour, so the cloud wins