That’s not quite true. A recipe can specify a URL to check for a new version, and the homebrew automation will periodically check it. If there’s a new version, it automates the version bump.
I used that for a package my company publishes, and neither we nor any other human AFAIK ever manually update it in homebrew, yet the newest version is always installable there.
Like what? I've never had problems related to PgBouncer, but apparently we have different use cases. I'd love to hear where the rough edges are so I can avoid them, or at least plan for them in advance.
I hated - hated - skeuomorphism. It’s not just the appearance, which I can take or leave, but also the idea of making software model the behavior of physical items. Frankly, the old Mac first party apps sucked for it. For example, for the longest time Calendar didn’t have an event list view, only a day or week or month view (IIRC; it’s been a while). Why? Because desk calendars don’t have an event list view, so neither should this. Or remember the old QuickTime Player where you had to adjust the volume by clicking the edge of a volume wheel and dragging it because that’s how an old portable TV worked.
Skeuomorphic icons can be pretty. Skeuomorphic UX sucks because it inherently brings physical world limitations to software where it’s unnecessary.
90% of the SPAs I use could be Django/Rails/Flask apps with no noticeable difference, other than that they'd be many times more responsive on slower devices.
"Old" SSR apps are mature, not obsolete. It's ridiculous when they're not considered even when they'd be the right tool for the job. I wouldn't try to clone Google Docs in Django, but something like Linear could 100% be modeled as a CRUD app with new page loads on click without a substantial loss of functionality from the user's POV. Not to mention built-in, automatic support for pervasive "deep linking", aka "linking", because you access a URL's contents by GETting and rendering that URL is just how it works.
I agree but you and I aren't the audience and I think experts in general should be a little more holistic when critiquing other people's choices. For any given problem there always exists a perspective from which your solution is over-engineered. People who (like us, I presume) understand processes, files, the command line, compilers, a computer language or two, a bit about computation theory (e.g. Turing, big-O, Knuth..) can get to a very broad swath of places along many different (often shorter!) paths. This is not where most people are starting from.
Speaking of Knuth, imagine being asked to "write a program to add two numbers" and using something like Python instead of assembler, or because really that's complex too, machine code. Do you think that the amount of housekeeping and computer activity is justified for adding two numbers? Objectively, it is not, its just that steady-state dominates the transient over time.
What models are you using? This looks incredibly similar to the outputs I get from Claude with default system prompt settings.
- First, take a look at the other articles. There are 30+ articles all with the same author published this week on topics from cooking to home appliances. Either he's extremely prolific or had help.
- There's so much click-baity LLM language: "The radial gradient: why edges undercook on every hob" "What “crispy” actually means: acoustic fracture mechanics" "What Actually Matters"
It's clear to me that at least some of this is LLM generated, so all of it might as well be.
I don’t think that’s right. Ghostty has its own TERM=xterm-ghostty value, which surely it wouldn’t need if it supported nothing more than ECMA-48/VT100.
That's usually only necessary for more questionable things that was niche for a reason, like file transfer, images, audio, and rediscovering the root cause of old security issues like this all over again: https://web.archive.org/web/20030302210517/http://www.digita...
reply