News for hackers. Stuff that matters.

grohan

grohan on Hacker News

Re:OccupySaaS on Sunday November 01, 2026 @10:28PM (#50332813)
spot on :)
Re:What TLA+ can and can't check on Thursday October 01, 2026 @10:07AM (#49921911)
I quite enjoyed Tony Hoare's essay - thanks for sharing. A lot of what he wrote has stood the test of time!

Show HN: Telos – build, run, and evolve software from declarative goals

Posted by grohan on Monday September 28, 2026 @11:50AM
from the show-hn dept.

https://usetelos.ai/blog/introducing-telos-cloud

Hi HN,I've been a heavy user of coding agents for over a year now, and while excited by the productivity gains - I've increasingly been frustrated at the process around it.- Prompts are imperative and ephemeral, and code is increasingly a black-box, so I lose track of intent over long horizons.- Pull requests, CI/CD, and even GitHub have become process bottlenecks and are increasingly losing...

( Read More... | no comments | on HN )
Re:Claude Opus 5.5 on Tuesday September 22, 2026 @01:56PM (#49805450)
Quite the load-bearing phrase!
Re:OpenSpec – A lightweight and configurable AI spec framework on Wednesday September 16, 2026 @09:51PM (#49735504)
how are people currently evaluating between the different spec writing frameworks out there?
Re:Tailwind Labs is joining Shopify on Wednesday September 09, 2026 @07:52PM (#49636305)
"load-bearing" - lol, nice easter egg. no one likes the AI-isms in writing!
Re:How well do agents use test/verification techniques? on Tuesday September 08, 2026 @08:19AM (#49609319)
Pretty good point - I would actually argue that the imperative nature of prompting takes some of the blame here too. Your initial goal gets "steered" by the agent system and follow-up messages take precedence.
Re:Ctoc: Cloc, but for Claude Token Counts on Tuesday February 10, 2026 @07:16PM (#46969043)
Here's a reverse engineer of the Claude token counter API as well as the model's vocabulary. Fast, offline, 96% accurate.
Re:llmfuse: A self-compressing filesystem backed by an LLM on Wednesday November 26, 2025 @11:19PM (#46065597)
Bellard has trained various models, so it may not be the specific 169M parameter LLM, but his Transformer-based `nncp` is indeed #1 on the "Large Text Compression Benchmark" [1], which correctly accounts for both the total size of compressed enwik9 + decompresser size (zipped).

There is no unfair advantage here. This was also achieved in the 2019-2021 period; it feels safe to say that Bellard could have likely pushed the frontier far further with modern compute/techniques.

[1] https://www.mattmahoney.net/dc/text.html

Re:Fire-Flyer File System from DeepSeek on Thursday February 27, 2025 @10:42PM (#43201470)
They appear to have Python bindings which seems reasonable from an API / usability perspective? https://github.com/deepseek-ai/smallpond

In terms of fast FUSE - also my first question, appears to be`io_uring` + FUSE :)

https://github.com/deepseek-ai/3FS/blob/main/src/lib/api/Usr...

Re:The Annual Cost of Technical Debt: $1.52T on Thursday January 25, 2024 @05:51PM (#39136540)
Impressive numbers. Does anyone have any read or anecdotes on how much a small/mid/large company loses from low quality of software/bad practices (or conversely profits from the opposite)?

Seems like a challenging metric to measure, but always been curious on what the numbers look like.