Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
sudb
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
6 ms
·
61.
▲
by
sudb
5mo ago
I think the idea of tracking intent in git commits is a great idea but it feels to me like this might be reducible to some prompts/extending git/pre-commit hooks?
62.
▲
by
sudb
5mo ago
Maybe some common complaints about Go are finally less of a problem in the current coding agent era - e.g. ecosystem weakness complaints and verbose error handling. Though TypeScript's type system is maybe still more powerful - and the
63.
▲
by
sudb
5mo ago
Yep! https://aws.amazon.com/blogs/aws/launching-s3-files-making-s...
64.
▲
by
sudb
5mo ago
Yes! I used keycloak for multitenant auth and it worked fine - a little dated but functional. Nowadays I'd probably stick to something like Clerk/BetterAuth/Supertokens.
65.
▲
by
sudb
5mo ago
I'm pretty unsurprised that the vision agent did worse. I'd be interested in a comparison between the different tools that now exist to let LLMs drive browsers (e.g. vercel's agent-browser, the relatively new dev-browser[1],
66.
▲
by
sudb
5mo ago
I didn't know that "Jamstack" was still a relevant term of art - I thought it'd gone the same way of LAMP.
67.
▲
by
sudb
5mo ago
Am I right in thinking that the picture provided with the blog post is the actual photo and not the inked engraving?
68.
▲
by
sudb
5mo ago
I've always found the pigtail selector to be the coolest part of the process which just uses geometry to "select" a crystal that grows in the correct orientation for the turbine blade.
69.
▲
by
sudb
5mo ago
There's most definitely many people involved in Banksy's art - but consensus seems to be forming that it's a man named Robin Gunningham [1]. 1. https://www.reuters.com/investigates/special-report/glo
70.
▲
by
sudb
5mo ago
Similarly I've often flip-flopped on the safety of public API endpoints that are "protected" by virtue of no sitemap + UUIDs in the URL path - I think the answer ultimately is that this is fine so long as there's no way
71.
▲
by
sudb
5mo ago
Is secretly rerouting reads/writes/edits of skills and memory any easier than just dumping the actual skills and memory files on disk at sandbox startup? Another benefit of moving the harness outside the sandbox is you get to avoi
72.
▲
by
sudb
5mo ago
At least not all the author's blog posts seem to be AI-written: https://danieltan.weblog.lol/2023/01/why-most-image-gen-star...
73.
▲
by
sudb
5mo ago
What's the underlying TTS model this is using?
74.
▲
by
sudb
5mo ago
Makes plenty of sense to upstream this (possibly makes more more than forking, although I suppose it's one way of gauging interest and implementation complexity).
75.
▲
by
sudb
5mo ago
So interestingly, I know of at least one application in a charity that deals with trafficking where grok was happy to do one-shot classification tasks where all other models refused to cooperate. I think there's a surprising number of
76.
▲
by
sudb
5mo ago
I am personally saddest to hear about the discontinuation of the Vision Pro - in a couple of generations there was a solid chance that it would be easy sell for me and/or other people who don't VR/AR game but probably would u
77.
▲
by
sudb
5mo ago
I think it's fine that it's minified, code-golf style, but yes I agree that this would be much more convincing and useful with an unminified source.
78.
▲
by
sudb
5mo ago
It feels unlikely to me that at least first few models at OpenAI/Anthropic were distilled from each other. If it is happening for newer model generations, I doubt either of them would ever publicly admit to this like Musk has done for
79.
▲
by
sudb
5mo ago
If you mean the self-driving part specifically, apparently Comma AI already does this: https://comma.ai/
80.
▲
by
sudb
5mo ago
I do think that many many labs do distill frontier models, but it's a shame that xAI also feel the need to do this - perhaps an indication or admission that in fact, they are not a frontier lab.
81.
▲
by
sudb
5mo ago
it's now my go-to for when I need to wrangle basically any text file manually - has handled everything I can throw at it (some of which has crashed other editors -looking at you Cursor/VSCode)
82.
▲
by
sudb
5mo ago
I think we call those buses, usually!
83.
▲
by
sudb
5mo ago
I suppose I primarily mean marketing - perhaps the most immediate concrete example I can think of is some sort of co-promotion alongside some mainstream vibe-coding tool that positions them as the git host of choice.
84.
▲
by
sudb
5mo ago
A tangential but interesting takeaway for me from this is that Harris Tweed was at some point in danger of dying out and that it was saved (?!) by now King Charles.
85.
▲
by
sudb
5mo ago
At this point maybe even Azure DevOps is an improvement
86.
▲
by
sudb
5mo ago
I'm not sure I understand why Poolside are training their own models at all - what's the perceived or real advantage of splitting up model training efforts into smaller companies and dividing up resources like this? Is it just to
87.
▲
by
sudb
5mo ago
As in, to present themselves as the new defacto git host, capitalizing on GitHub's actual + perceived lack of reliability
88.
▲
by
sudb
5mo ago
I've been trying to figure out what the long term play is here - is it an angling for a frontier lab acquisition? Or does open-sourcing put Warp in the same sort of category as OpenCode - where charging for LLM tokens becomes the main
89.
▲
by
sudb
5mo ago
I'm very interested in where ghostty ends up - I wonder if they'll follow Zig to Codeberg? It does seem like it might, in general, be a very opportune time for GitLab (or another host) to publicly step up! There seems to be a lot
90.
▲
by
sudb
6mo ago
Anecdotally, conversion - from free to trial, trial to paid, one-off purchases, etc.
More ›