sorry to hijack the thread. Really cool post. How long did the whole exercise including porting zlib to lean take?

i have a hard real time system that i would love to try this on, but that's a lot of tools to learn and unclear how to model distributed systems in lean.

also, please add rss so i could subscribe to your blog

Lean-zip was not my project but one by others in the lean community. I'm not sure about the methodological details of their process - you might want to check with the original lean-zip authors (https://github.com/kim-em/lean-zip)

Worked on that phone. I diagnosed if music player cache was working by putting back of phone to my ear and listening for seeking clicks from hdd

Nice! I loved that phone, was one of my favourites. One of the only, if not the only phone I ever owned that had a metal shell that I an recall.

I had most of the N range, and was particularly interested in music ones, N95 was love/hate because the music button/reverse slide was so slow sometimes, and generally it just wasn't as good as N91 for music listening with its proper headphone jack placement, and always accessible controls.

What kind of magic did that HDD have that it could be thrown around like a phone typically is without the issues we would see if we'd handled a laptop with HDD the same way?

Had an accelerometer that turned off hard drive when motion detected. High end ibm thinkpads had that too. Turns out if seek head is parked properly, its fairly robust

N91 also had a ridiculously high quality DAC that beat pants off iPods of that gen.

I think I remember reading something about that at the time, the accelerometer thing, maybe it was in the manual.

Sounds quality was great and I loved the dock too.

yeah all phones were supposed to survive 2m drop test on concrete and still be able to make a call

This is why aider only edits files via git commits

your patch queue approach is very clever. Solves a huge tech debt poblem with llm code gen. Should work with jujitsu too probably.

Would be curious to see more about how you save tokens with lua too.

Do you blog?

Thanks for your interest in this work - I do not blog(maybe I should?) but i have posted a bit more on X about this work.

- A bit more on mail mode https://x.com/hsaliak/status/2020022329154420830

- on the Lua integration https://x.com/hsaliak/status/2022911468262350976 (I've since disabled the recursion, not every code file is long and it seems simpler to not do it), but the rest of it is still there

- hotwords for skill activation https://x.com/hsaliak/status/2024322170353037788

Also /review and /feedback. /feedback (the non code version) opens up the LLM's last response in an editor so you can give line by line comments. Inspired by "not top posting" from mailing lists.

I quit x so cant read beyond toplevel links. I subscribed to your tool on github, would appreciate blog-posts-in-release notes to keep up with future developments. Will try the tool. Rare to find something new among ai hype, thank you.

Fair enough. I'll find a way to publish some of this. I try to cover most of the information in the docs/ folder, and keep it up to date. Blog posts in release notes is a good idea!

The db that was the cause of redhat rpm db corrupting itself

hash collision limitation for keys is the most questionable part of design. Usually thats handled by forcing key lookup to verify that what you looked up matches what you tried to lookup. Resolving this perf hit is probably doable by having an extra table of conflicting hashes

(author here)

The documentation page is out of date, the format now resolves collisions through quadratic probing.

I love that age lets one reuse ssh identities and thus identity sharing systems. The single most useful thing I ever wrote was a tool to sync github identities with age. https://github.com/tarasglek/github-to-sops

This way you get git for change tracking on your secrets and who-has-access-to-secrets and key rotation and this can be trivially expanded to other forges.

Its easy to introduce age this way into any modern project whereas gpg would-ve been a non started on most teams I worked on.

disclaimer: this was mostly vibe-coded because I really did not want to work on this and wasnt sure if teammates would adopt it. Then it just worked, so stayed ugly inside

my $100 mobiscribe lets me read HN. I only read online on my eink devices. LCD is much less enjoyable

To make it open source in the fullest sense one needs to document what youve done. This esp repo could use some details on what protocols the hardware speaks, sequence diagrams, auth, etc. I doubt you running webrtc on esp

WebRTC is running on the esp32! Library is libpeer[0]

[0] https://github.com/sepfy/libpeer

i am still confused what their software stack is, they dont use ceph but bought netapp, so they use nfs?

The NetApps are just disk shelves, can plug it into a SAS controller and use whatever software stack you please.

but they have multiple head nodes, so its some distributed setup or just active/passive type thing?

We have a custom barebones solution that uses a hashring to route the files!

I'm guessing the client software (outside the dc) is responsible for enumerating all the nodes which all get their own IP.

I think each rack is one head node and several disk shelves (10?). No dual headed shelves.