Hacker Newsnew | past | comments | ask | show | jobs | submit | reneberlin's commentslogin

I fully agree. The psychological burden is also high, what makes the maintainer feel miserable over time.

A few weeks ago i saw a repo in Github that had closed the issue-board and explained the users: to move on to the discussions.

If the discussion leads to the insight, that this is worth a feature, the maintainer will implement it.

I think the same goes for PRs. If you think you have made something cool, then discuss it and offer your implementation and the maintainer makes it fit with you to integrate.

On the other side, if you want a feature it just doesnt have, simply forking it and updating from the source-repo is fair enough.


bash? :) iframes, a little dashboard, or nowadays you use something like mcp-browser-use, chrome-devtools-mcp with your llm of choice and write skills for that.

If it has APIs, you dont even need that browser-engine at all.

then you start the complete report with:

/morning


Yeah I haven't looked into dashboarding, as i always thought it took too long and was too limited. But i will try now thanks :)


I agree fully, up until the end of the sentence - my take is: everything else should be Webassembly/wasm.

If it's not a page, it's an app and that has to be wasm, because you are not allowed to treat the DOM of the browser so badly and unnecessarily bad, that it chokes under the pressure to render 1 view.

Example: https://www.egui.rs/#demo


I know that error since, dunno, i guess: 1995. No other OS has this unprofessional and annoying bug. And no other OS vendor managed it to drag a bug like this for 30 years. None that i know of.

Don't forget there are younger people, thats maybe new to them, or they seek the error on their own side. Just unbelievable to read that 30 years later.


Yes, of course you can! It's just preferrable to use the fingers instead of the hand.

Instead of letting the agent do it for you, you can prompt it to describe it and let you do the job.


Can you elaborate?


compare it to jp2a / ascii-image-converter In ascii they perform very well without ML, of course. You could set up a training based on that.


btw I optimized it for edge detection, but it pretty much ignores luminosity. So it performs better at rendering text / line art, but worse at pictures without clear boundaries


Thanks. Very helpful.


I love the idea, but most of the alternatives give up because it will be abused fraudulently soon and you might be liable in the face of the law because of your ip-adress and domain.

I really don't want to kill the fun, but there are risks that are real.


The last one i knew was: https://serveo.net/ and it seems to be gone. The decline came in waves. First he made a full-stop, then put in some mittigations and gave up in the end.

https://github.com/milio48/serveo

It think you could ask politely for his ride on GH? milio48


Thanks for the feedback! Any tips for how this might be remedied? Limit free tier to 1 hour per tunnel? Clear ToS and liability terms? Some form of automated abuse monitoring? I don't imagine it would be any more liable than ngrok itself. But yeah, you make some great points!


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: