solarkraft

5 ms·

1.
by solarkraft 7d ago
Automattic also bought Tumblr and Pocket Casts, publishing the code of the latter. I used to have a subscription just to support them.
2.
by solarkraft 7d ago
He still contributed a lot to the world during the good times.
3.
by solarkraft 7d ago
I’m a bit relieved. Only knowing it superficially through its larger public actions, Automattic used to be one of the “good guys” companies to me. I envisioned myself applying there at some point. This has obviously changed over the last 1-
4.
by solarkraft 7d ago
27 is really a lot better, even on an iPhone 13. Yet that’s still just repairing what they previously broke. I still can’t listen to music AND listen to a voice memo or a video clip.
5.
by solarkraft 7d ago
I’d expect their chips to support it by now, but LE Audio, which Auracast is a feature of, requires isochronous streams, a feature of Bluetooth 5.2.
6.
by solarkraft 7d ago
Back when BLE Audio was first announced, I thought Apple would champion it because it finally extends quality limitations making the microphone quality terrible. I guess they’re content with doing their proprietary thing.
7.
by solarkraft 7d ago
I agree that a CSS library is not really worth a big business. But nobody has woken up: AI makes writing tailwind classes more convenient than ever and that’s what people seem to do.
8.
by solarkraft 7d ago
Prefill is generally faster than generation, but not by much on older Mac processors. I get around 70-60 tps in prefill on my M1 Max for Muse Glimmer (not sure about the generation speed, probably between 15 and 30). They allegedly improved
9.
by solarkraft 8d ago
I don’t even open twitter/x links anymore because the page is so unusable.
10.
by solarkraft 8d ago
Grok is made by Elon Musk and hence fit for neither
11.
by solarkraft 8d ago
I’m a big fan of the IA and have donated before. I would really appreciate it if they spent more resources on improving the site’s UX and fewer on picking pointless and expensive copyright fights.
12.
by solarkraft 9d ago
Good for X to be somewhat up again, I guess
13.
by solarkraft 12d ago
It’s obviously a trick.
14.
by solarkraft 14d ago
I use OpenCode with my ChatGPT subscription, which is officially supported on both sides and a damn good deal (OpenCode Go is great too). OpenCode has “providers” for many (many!) other services, but these are almost all unofficial and agai
15.
by solarkraft 14d ago
Perfect, I don’t have to go on this tangent on my own! Apple’s dictation feature has great theoretical UX and an implementation so bad that it practically does not exist.
16.
by solarkraft 14d ago
> These are not market personas. They are architecture tests. Indeed.
17.
by solarkraft 15d ago
What’s cynical about this? Everybody is modernizing their codebases for agentic development.
18.
by solarkraft 15d ago
> Under a playlist you get a list of suggested songs, the UI element looks like a "regular song", but unlike every other "song" element the user has been trained to recognize, the "swipe right to queue" does
19.
by solarkraft 15d ago
If going through a strongly designed process with quality gates is vibe coding, I guess so. I thought the term was reserved fir “claude, make this happen” style development and the term for the more professional approach was “agentic develo
20.
by solarkraft 15d ago
You’d have to do the relevant R&D and maintain the product.
21.
by solarkraft 16d ago
My guess is that they will continue to care as much as they currently do, get a symbolic fine and carry on unchanged.
22.
by solarkraft 16d ago
My OpenCode setup has most of this. Automatic resumption is kind of cool. Could you go into the the tooling for long-lived loops?
23.
by solarkraft 16d ago
To me it’s kind of the dumbest thing about it. “Talk to claude from anywhere” - huh, like I already do? Why can’t I have an extra app? How do I manage my sessions? I (as somebody also DIYing) consider the defining feature of a “claw” to be
24.
by solarkraft 18d ago
NGL: I don’t really have a good way to find out right now. It also doesn’t matter that much because the way the models use the tokes varies a lot. Qwen 3.8 is known for overthinking while Muse Glimmer may be a little slower per token, but i
25.
by solarkraft 19d ago
So far I don’t regret buying an M1 Max device with 32Gb of RAM. The models available for it keep getting better (running just about okay for interactive use) and 400 GB/s of bandwidth is still considered a lot. The models are currently
26.
by solarkraft 20d ago
Logseq does this too - or used to. They found that for the rich operations they want to support (live collaboration, richer data), persisting the data in a graph structure makes more sense to them. It probably adds some conveniences, but th
27.
by solarkraft 20d ago
Simplicity. It’s a bunch of files, trivial to edit, version and develop against. This is important for things sometimes intended to stay relevant for a long time. I’d flip the question towards the benefits more complicated solutions give yo
28.
by solarkraft 20d ago
I do this and plan to intensify it. With one important feature: A “summary” field that can act as a preview when listing your files so your agent’s context doesn’t get jammed.
29.
by solarkraft 21d ago
Journalist discovers people using analogies ... The author realizes this later in the article, but still seems to conjure up something about technology being special in this regard. It's not. I've got an ace up my sleeve. I hit a
30.
by solarkraft 22d ago
This is ultimately X’s problem. Remove the ways to consume its content, remove its relevance. If at all, people should only fully quote content or post screenshots now.

More ›