Hacker News: olejorgenb

New comment by olejorgenb in "Anthropic downgraded cache TTL on March 6th"

olejorgenb — Sun, 12 Apr 2026 12:44:02 +0000

Maybe they should fix bugs like this then https://github.com/anthropics/claude-code/issues/17979#issue... ...

New comment by olejorgenb in "Issue: Claude Code is unusable for complex engineering tasks with Feb updates"

olejorgenb — Thu, 09 Apr 2026 14:40:51 +0000

I see. If that's just hashes and not encrypted content I can't see how they can resume old sessions properly. IIRC they have a 30 days retention policy and surely the thinking traces must be considered data. Wonder how this works with the zero-retention enterprise plans...

New comment by olejorgenb in "Git commands I run before reading any code"

olejorgenb — Wed, 08 Apr 2026 21:45:48 +0000

git bisect --first-parent

New comment by olejorgenb in "Issue: Claude Code is unusable for complex engineering tasks with Feb updates"

olejorgenb — Wed, 08 Apr 2026 10:36:09 +0000

So this means I can not resume a session older than 30 days properly?

New comment by olejorgenb in "Embarrassingly simple self-distillation improves code generation"

olejorgenb — Sat, 04 Apr 2026 21:47:28 +0000

> I wonder if there is a more general solution that can make models spend more compute on making important choices, while making generation of the "obvious" tokens cheaper and faster.

I think speculative decoding count as a (perhaps crude) way implementing this?

New comment by olejorgenb in "F-15E jet shot down over Iran"

olejorgenb — Fri, 03 Apr 2026 23:48:09 +0000

Source of that image though.. ?

New comment by olejorgenb in "Claude Code Unpacked : A visual guide"

olejorgenb — Wed, 01 Apr 2026 13:42:13 +0000

The tools was mostly already known, no? (I wish they had a "present" tool which allowed to model to copy-paste from files/context/etc. showing the user some content without forcing it through the model)

New comment by olejorgenb in "[dead]"

olejorgenb — Tue, 31 Mar 2026 15:12:48 +0000

> This is a production-grade agentic system that happens to live in your terminal.

You read the code?

> The ink/ directory — roughly 50 files — is not the popular npm ink package. Anthropic built their own React-based terminal rendering engine from scratch.

Interesting

New comment by olejorgenb in "If you don't opt out by Apr 24 GitHub will train on your private repos"

olejorgenb — Fri, 27 Mar 2026 22:11:46 +0000

To be fair, they display it reasonable prominently in GitHub when you are logged in. Given that, I feel the post title fall under the click bait category. I was fully aware of the Co-pilot opt-out change, but still clicked due the phrasing of the title.

New comment by olejorgenb in "Shell Tricks That Make Life Easier (and Save Your Sanity)"

olejorgenb — Thu, 26 Mar 2026 17:26:53 +0000

Do yourself a favor and upgrade your history search with fzf shell integration (or similar): https://youtu.be/u-qLj4YBry0?t=223 / https://junegunn.github.io/fzf/shell-integration/

New comment by olejorgenb in "Opera: Rewind The Web to 1996 (Opera at 30)"

olejorgenb — Tue, 24 Mar 2026 12:33:24 +0000

Opera was by far the best browser for a while for sure. Sad they couldn't keep up :/

New comment by olejorgenb in "Cursor Composer 2 is just Kimi K2.5 with RL"

olejorgenb — Fri, 20 Mar 2026 21:14:00 +0000

Well, the UI as a whole is ok to me (except the parts which is way too volatile). I was talking about the UX of the autocomplete model. The model are very often spot on and fast, but it's impossible to properly configure it to be less in your face. Making it basically useless for day-to-day development.

New comment by olejorgenb in "Too Much Color"

olejorgenb — Fri, 20 Mar 2026 11:53:43 +0000

My impression is that they are a compression expert, not a color expert. Make sense they chose uniform flat colors :D

New comment by olejorgenb in "Cursor Composer 2 is just Kimi K2.5 with RL"

olejorgenb — Fri, 20 Mar 2026 11:35:36 +0000

The model is great. The UX is ~~horrible~~ annoying...

New comment by olejorgenb in "Cursor Composer 2 is just Kimi K2.5 with RL"

olejorgenb — Fri, 20 Mar 2026 10:51:07 +0000

To be fair, is "with RL", "just"?

They should have disclosed it though. If they didn't it's a bad look for sure.

New comment by olejorgenb in "A Decade of Slug"

olejorgenb — Tue, 17 Mar 2026 21:28:03 +0000

> I was granted a patent for the Slug algorithm in 2019, and I legally have exclusive rights to it until the year 2038. But I think that’s too long. The patent has already served its purpose well, and I believe that holding on to it any longer benefits nobody. Therefore, effective today, I am permanently and irrevocably dedicating the Slug patent to the public domain.

New comment by olejorgenb in "Kagi Small Web"

olejorgenb — Tue, 17 Mar 2026 18:06:54 +0000

As I'm not familiar with how Kagi is "pushing ai tools" this is mostly a comment on the framing of your question.

Are you really saying that a company specializing in search - natural language oriented at its core - should not make use of the biggest technological revolution for processing natural language?

New comment by olejorgenb in "Kagi Small Web"

olejorgenb — Tue, 17 Mar 2026 17:42:16 +0000

I would trust Firefox if they made a version which did the indexing locally. I think I'd trust chrome as well as long as it was implemented locally.

New comment by olejorgenb in "Kagi Small Web"

olejorgenb — Tue, 17 Mar 2026 17:12:31 +0000

Ah :wooosh:, :D

New comment by olejorgenb in "Kagi Small Web"

olejorgenb — Tue, 17 Mar 2026 14:59:07 +0000

I'm 80%+ sure the claim is BS