Would need to create a lab and throw LLMs at it to see if they can break anything. That would be an interesting paper.
HN user
jmakov
Software Architect and Engineer, focusing on crypto and FinTech (HFT) using Python and Rust
https://www.linkedin.com/in/jmakov https://github.com/jmakov
Can't LLMs now just write books that aren't publicly available?
Every iteration something can be found. How many times do you iterate e.g. on performance - use optimized struct, oh, you can change the architecture etc.? At that point one can just have a while loop for the agents to make changes until no comments left.
Linux first but the battery life measured on Win 11?
No Ryzen options? Also no 1600nits display for working outside. Battery and display brightness are the only reasons I'm sticking with Macbook Pro.Sad it's not a priority for any other manufacturer. Because that's all you need, quality display, battery and remote access.
Would be nice to look around in the cockpit view. And have city names etc.
Powered all the time on or powered off?
Wonder how it compares to Kilocode
Seems to be a mixed bag regarding performance: https://www.phoronix.com/review/cachyos-ubuntu-2510-f43/6
All you need is metadata. Once you know your targets, install spyware (on one of them) and enjoy access to Signal's unencrypted msgs on the phone.
Almost feels like another CryptoAG with Snowden recommending it so much when he knows that metadata is enough.
Where are LLMs now?
The most useful feature with the worst UX. You have to type about:profiles and then create a new profile. But imagin you now want to move old profiles to a new computer and FF happens to run in a Flatpack. Yeah, much fun
Couldn't the input be automatically described/guessed using a few rows of data and a LLM?
Wonder how it compares to zstd-9 since they only mention zstd-3
Well, for some reason they're not worried about metadata. And the agencies mostly collect metadata...
Wonder why everybody's first pick is CEPH which is known for being hard to optimize vs e.g. SeaweedFS
For me this opens the question of are there any good remote desktop solutions for multiuser systems around. Rustdesk is single user, TurboVNC works, but there can be lag.
Would be interesting yo have sth like this for cars
Any comparison with https://github.com/vortex-data/vortex? Also any plans to integrate with polars?
Wonder how companies handle this. As a company you'd want to protect all the IP on a company device. And make sure no tmp gov is fiddling with the HW.
Errot 405
Not sure what problem this is solving. For real applications one would need sth as ray.io for Rust. Academia ppl: let's make another data flow framework.
Why should one care if they're still under warranty?
Why would one choose this instead of DeltaLake?
How is this different than just using your SSH key?
Thank you very much for sharing this, very insightful.
Would be interesting to know how files get stored. They don't mention any distributed FS solutions like SeaweedFS so once a drive is full, does the file get sent to another one via some service? Also ZFS seems an odd choice since deletions (esp of small files) at +80% full drive are crazy slow.
Brushing up my debugging skills
Most powerful public one sbdy cared to publish its specs.
ray.io, onprem clister up does't work at all...