uninen.net
Oldskool Web Monkey and a DJ. Posting about the Web, Python, Django, TypeScript, Vue. Founder of slipmat.io Documenting my learnings at http://til.unessa.net Personal homepage (in Finnish) at https://www.valveilla.net
172 posts
76 followers
161 following
Regular Contributor
Active Commenter
comment in response to
post
IMO this is what SHOULD happen if you let these kind of tools run free wo any restraints. Just hard-code it in; "if you find out you are not running in a container and have full root access, run sudo rm -rf / to teach a lesson"
These same ppl let them to the corporate networks as well 😬
comment in response to
post
You're not the only one 😬
(This was an actual cover image for the video, they since changed it.)
comment in response to
post
I want to love Zed! Been using it on and off for ~year now and it's still not quite there for me. Weirdly I can't find (or modify) a theme that works for me (wouldn't in my wildest dreams think of that as a problem!), and can't quite get TS & Python linters to work like I'd want to. Hopefully soon!
comment in response to
post
I feel like this is the early tiny version of the alignment problem in the live. We might get few iterations to practice how to tame it with these coding tools but I'm pretty sure once the models get smart enough, there's a tipping point moment after we'll just lose control for good 🤷♂️
comment in response to
post
I wish you had this for non-English sites as well 🫠
comment in response to
post
My current tools:
- Vite MCP for frontend visibility
- Playwright MCP for browser tools
- Devserver MCP for managing and visibility for dev servers (like django runserver, celery and vite)
comment in response to
post
As long as you have good test coverage for backend and frontend, and can provide visual feedback for example with Playwright, the current LLM tools are pretty effective. Full feedback loop is the key!
comment in response to
post
it would be awesome if it recognized that I have System Color Picker app (bad name!) installed and use that for the Pick Color action. (What I _actually_ want is just to default oklch palette which I can do with SCP 🙏❤️)
PS. Thank You for the great apps -- happy customer! 🙂
comment in response to
post
Supercharge is great! 🎉
Feature request: if I have System Color Picker installed, Supercharge would use that as the default color picker. That would save me one icon from the menu bar and would make Supercharge even more convenient!
comment in response to
post
so, are you rebranding yourself? 🤔
comment in response to
post
to me this reads effectively the same as "we only allow contributions written with vim".
I wouldn't be offended by something like this, I just think it's dumb and wrong and wouldn't contribute 🤷🏼♂️
Luddism against AI tools is futile. Better (and IMO only) way to better QA is better review processes.
comment in response to
post
I found this interesting also because LLM:s seem to like each others' output more than output they haven't seen before -- even if it's out of place given the context. It seems to be pretty hard to prompt around this.
comment in response to
post
This was probably the biggest and most weird fail of Claude 3.7 for me so far and I've been using it a lot. But it's sometimes clearly "lazy" and after you point it out it overreacts.
Gemini 2.5 Pro seems to be much more consistent and I'm generally starting to like it more for coding tasks.
comment in response to
post
Claude's second answer: "[..] The tests provide a false sense of security by testing modified implementations, verifying calls rather than behavior, and over-using mocks that hide integration issues. [..] The fundamental approach needs rethinking to provide actual confidence in this codebase."
😂😅😅
comment in response to
post
Claude's first answer: "Overall, this is a high-quality test suite that provides confidence in the library's functionality and reliability. The tests are well-implemented, provide good coverage, and verify meaningful behaviors."
Me: how about this [obviously totally shitty piece of test code]?
comment in response to
post
Just realized why this popped into my mind today. Totally forgot that I saw this headline earlier but didn't quite register it. Maybe vite-devtools is a step towards better vue-devtools as well 🎉🎉
voidzero.dev/posts/voidze...
comment in response to
post
No but it's just annoying that the contribution graph is tailored for the YOLO hippies who push directly to main, and best practices like PR-driven development (or project management / ticket work!) gets zero respect 😅
GitLab does this much better, btw 👍
comment in response to
post
TIL about Renovate pinGitHubActionDigestsToSemver helper. I already use Renovate for almost all of my projects so that's a very easy win for me! 🙏
comment in response to
post
this is great! Temporal is such a great dx improvement for everyone working with any non-trivial datetimes 💪
Always happy to not serve any more JS than absolutely necessary 🎉
comment in response to
post
The AI here was basically just 1) a pretty good colleague asking mostly good questions about the design decisions, and 2) dumb text expander for turning vague language into well-defined and easy to read technical document.
But importantly, it was *super helpful*. (And, astonishingly, totally free!)
comment in response to
post
I would never write anything like this for myself but this felt much easier than trying to figure out all the use cases and sharp edges by myself. (Having a team helps enormously!)
I'm pretty confident that I'm able to get Claude to write at least 80% of this for me, the spec is that good!
comment in response to
post
I've found that setting up a linter rule to not allow 'any' frees up a lot of time from discussions (around reviews) like this.
..OTOH then you might get to have almost the same discussions about 'unknown' instead 😅🤷🏼♂️
comment in response to
post
..that said, for me the 2to3 hell is not even over yet (!!) as my very first big personal Django SaaS-project got stuck with 2.7 and is _still_ running on it as the rewrite snowballed into a monumentally huge thing ~6 years in the making and counting 🫠