HN user

paweladamczuk

246 karma
Posts1
Comments113
View on HN
Transcribe.cpp 3 days ago

I've been using this one for a week for local transcription, working pretty well so far

Qwen 3.8 3 days ago

Check cache hits in your logs. You can use Openrouter or pi config to pin providers with best cache hit rates (or disable ones with the worst).

I use Openrouter for everything except Deepseek. For Deepseek I use their API directly.

I once plugged in a Logitech keyboard on a fresh system and got a colorful branded popup covering several inches of screen real estate in the bottom right corner. It was urging me to download some Logitech software.

As far as I know, the source of the graphics was not the unifying receiver that I plugged in the USB port, and the notification was not using any OS API meant for hardware to be avle to prompt the user for additional download. It was a Logitech-built DLL shipped and loaded by the operating system as part of some default driver for the Logitech keyboard.

I wonder if the functionality od Index 01 will also be available via the Pebble watch? I mean, it would be logical to allow that, all it needs is a separate app for the watch. But I didn't find any clear confirmation of this anywhere.

When I was setting up Wireguard server at home, the last step (after testing the setup) was the LLM giving me a set of commands to run manually to regenerate the credentials.

Immich 3.0 20 days ago

That's what I do. I have Wireguard connected on my iPhone at all times.

LLMs became so good that I don't trust a codebase like Immich to have ports to my server exposed publicly.

I put everything behind Wireguard to limit the number of lines of code that might bring down my setup.

To add to the sibling comment, you are also ignoring the fact that in 1999 nobody had those conveniences, everybody was on equal ground. In 2026, if you handicap yourself by rejecting those "conveniences", you will be met by friction at every step - lower productivity at work, impatient looks from your family members etc.

That's similar to the direction I went with my PC. It's a server that sits in a datacenter. It is wireguard protected and has SSH access for general stuff, copyparty for file access, webtop in a container for graphical tasks like audio editing, software like Navidrome for music and Immich for photos.

I could just call it a "home" lab server. But I actually use it as a general purpose computer, not just a server.

I don't think there can be tool calls inside the obfuscated reasoning blocks. I mean, in order for those function calls to be evaluated client-side, that thinking stream would have to be decrypted on the client side at some point, which would defeat the purpose of obfuscating it the way they do.

If you mean the function calls might happen server side, there is nothing preventing the server from doing it and hiding it from you as long as you are using an API for inference.

What about the cybersecurity aspect of bespoke software?

A cybersecurity research company can now spend a small fortune on finding zero days in iOS because of the amount of people that use it. It basically guarantees there will be clients like government agencies willing to pay through the nose for the exploits.

Software made for one might disrupt this business model.

Do you have a way to use voice on your smartphone?

I use a whisper wrapper (also "built for one") for Windows and connect through SSH but it only works well enough because of the laptop's ARC graphics card. I wish I could do that when connecting to SSH from my smartphone.

DeepSeek v4 3 months ago

these thing just take a while

Oh, so people might be forced to give back the AI earnings? Should I be worried about the last year's capital gains on my portfolio?

I was recently running Copilot CLI in a sandbox on autopilot mode and it kept overriding git config to put only "GitHub Copilot" as commit author instead of my name. Strongly worded instructions weren't helping, I had to resort to the permission system to change this behavior.

I wonder if this is consistent with their terms of service. I mean, maybe they DO take all the responsibility for the code I generate and push in this manner?

This article and the articles linked in it only provide a selection of works entering public domain in 2026. Does anyone know of a database or list of works so that I can see all of them? Other than the Wikipedia article that only has a list of names.

Beets with this flag works relatively well for me when wrapped in a shell script, but I have spent a lot of effort trying to make it actually non-interactive.

I'm still not totally sure I got all the certainty thresholds right, I still sometimes get unexpected behavior when using Beets this way, so I agree that the non-interactive experience is lacking.

Internet in its current form, where I can theoretically ping any web server on earth from my bedroom, doesn't seem sustainable. I think it will have to end at some point.

I can't fully articulate it but I feel like there is some game theory aspect of the current design that's just not compatible with the reality.