Hacker News: summarity

New comment by summarity in "The AI Situation in Software Development"

summarity — Sat, 15 Aug 2026 16:45:18 +0000

Keeping a simple log of "accepted/rejected" avenues is basically all that's needed, maybe with a root style guide in the project. I've got several multi-day sessions in 5.6 Sol running without going off the rails in terms of complexity. After a while in this loop it actually starts to remind/berate itself to keep things straight-forward.

New comment by summarity in "Racket v9.3"

summarity — Sat, 15 Aug 2026 05:43:50 +0000

Thank you for Herbie btw, it continues to be one of the most used tools in my personal projects (all 3D math and rendering)

New comment by summarity in "Mistral Patent for “Code implemented tool calls”"

summarity — Mon, 10 Aug 2026 18:01:17 +0000

I agree in general, but can think of at least one counterpoint: https://terathon.com/blog/decade-slug.html

Actually novel implementation is protected, paid the author's bills, and was dedicated to the public domain recently - no massive corp involved.

Rich Text Editing Innovations in Bike

summarity — Mon, 10 Aug 2026 11:21:28 +0000

Article URL: https://www.hogbaysoftware.com/posts/bike-rich-text/

Comments URL: https://news.ycombinator.com/item?id=49242206

Points: 2

# Comments: 0

New comment by summarity in "F*: A general-purpose proof-oriented programming language"

summarity — Sun, 02 Aug 2026 19:52:34 +0000

What? There’s literally a completely interactive book linked right from the home page.

New comment by summarity in "Agent Skill to Force Docs in ASD-STE100 Simplified Technical English"

summarity — Fri, 31 Jul 2026 01:55:15 +0000

"Test B is an alternative to test A."

could mean there's a Test B and a Test A, and they're interchangeable.

Or:

You can run Test B to confirm A works, that is "to test A".

Again the stated goal to clear documentation for non-native or limited-exposure speakers. This doesn't pass that test.

New comment by summarity in "Agent Skill to Force Docs in ASD-STE100 Simplified Technical English"

summarity — Thu, 30 Jul 2026 22:50:27 +0000

Gotta love how the very first example in Issue 9 of the standard is already self-defeating:

> "Test" is an approved noun, but not an approved verb.

> STE: Test B is an alternative to test A.

So much for clear - unless you know the STE specific rule, the sentence is unambiguously ambiguous.

Direct access btw since the official site gates downloads with a Google form: https://www.asd-ste100.org/assets/files/ASD-STE100_ISSUE9.pd...

New comment by summarity in "Blender 5.2 LTS"

summarity — Mon, 20 Jul 2026 09:44:05 +0000

To answer the question at the top, yes there are countless suites that are instantly resumable even after months away. I used to teach Blender (v3) and I think I could find my way around but the overhead isn’t worth it. I’m also though in a privileged position to be able to afford more, smaller commercial tools.

CAD: MoI (cool kids nowadays use Plasticity, but that is taking on similar feature creep - in any case I own and use both)

SubD modelling: nothing has ever beat Silo. And silo still gets updates, but the workflow is the same. Cheetah3D is also in the same boat.

Rendering (I do products, not character/CGI): Maverick. Fastest path tracer in the world, perpetual license, rivals Vray for product viz. if you’re currently using Keyshot - switch now.

Texturing: 3Dcoat or Marmoset Toolbag 5

What most of these have in common is either the workflow is super intuitive, or you have at least 15 years of training material readily accessible that still applies to the software today as it did back then.

Overall it’s a choice between a kitchen sink pipeline (Blender, Rhino) vs a specialised one. If I were to teach again, I would still use Blender in the classroom, and I fully support the project. My brain just works better with smaller, focussed tools.

New comment by summarity in "Blender 5.2 LTS"

summarity — Mon, 20 Jul 2026 09:35:22 +0000

The question is why don’t they integrate cycles? I mean I have a random app for home planning on macOS and even that comes with Cycles as the rendering engine. Can’t be that hard.

New comment by summarity in "OpenAI loses trademark dispute at EU court"

summarity — Wed, 15 Jul 2026 16:00:41 +0000

Key difference between the trademark systems here: in the EU system you don’t get a trademark by trading with a specific name and it then being recognized. It’s the other way around: the name must be unique, not confusing, and highly specific. It’s actually irrelevant whether a product exists or is traded at all.

Having gone through the process and gotten both approvals and rejections, the line is pretty clear.

New comment by summarity in "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor"

summarity — Mon, 13 Jul 2026 17:11:40 +0000

Vs Voxtral would be a better comparison. No other model, open or closed, has been able to hit such a low AER (Acronym Error Rate ;)) for my meeting transcripts. Seems to understand/infer all the technobabble I use at work. Never have to edit anything. Whisper was catastrophically bad.

New comment by summarity in "Kimi K2.7 Code is generally available in GitHub Copilot"

summarity — Thu, 02 Jul 2026 10:14:45 +0000

It has supported custom, local, any BYOM for quite a while.

I work at GitHub but even then I often use OpenRouter models in the CLI and Copilot App

LLMs use "safety" specific neuron layers to identify vulnerabilities in code

summarity — Wed, 24 Jun 2026 22:12:25 +0000

Article URL: https://arxiv.org/abs/2605.29901

Comments URL: https://news.ycombinator.com/item?id=48666231

Points: 5

# Comments: 3

New comment by summarity in "We’re making Bunny DNS free"

summarity — Wed, 24 Jun 2026 10:19:52 +0000

Their DNS is also scriptable, it’s not just a name server

New comment by summarity in "Remembering Planet Source Code: Sharing Code Before GitHub Made It Easy"

summarity — Tue, 12 May 2026 13:31:51 +0000

You can still download VS6 from Microsoft and clone a repo from there, and chances it’ll compile and run are higher than JS project that’s two weeks old

DeepL lays off 25% of its workforce

summarity — Fri, 08 May 2026 14:45:43 +0000

Article URL: https://www.heise.de/en/news/Cologne-AI-translator-DeepL-lays-off-a-quarter-of-its-workforce-11286250.html

Comments URL: https://news.ycombinator.com/item?id=48064003

Points: 9

# Comments: 0

New comment by summarity in "Claude Code Routines"

summarity — Tue, 14 Apr 2026 17:31:17 +0000

If you’re trying this for automating things on GitHub, also take a look at Agentic Workflows: https://github.github.com/gh-aw/

They support much of the same triggers and come with many additional security controls out of the box

New comment by summarity in "NimConf 2026: Dates Announced, Registrations Open"

summarity — Tue, 14 Apr 2026 14:42:33 +0000

It’s growing but not a lot, I have some data here: https://pierretempel.com/p/nim-usage-on-github

Most code I write is still Nim though.

Copilot CLI can now ask a second model to critique the first

summarity — Tue, 07 Apr 2026 13:00:15 +0000

Article URL: https://github.blog/ai-and-ml/github-copilot/github-copilot-cli-combines-model-families-for-a-second-opinion/

Comments URL: https://news.ycombinator.com/item?id=47674643

Points: 3

# Comments: 0

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

summarity — Mon, 06 Apr 2026 16:11:49 +0000

Not claude code specific, but I've been noticing this on Opus 4.6 models through Copilot and others as well. Whenever the phrase "simplest fix" appears, it's time to pull the emergency break. This has gotten much, much worse over the past few weeks. It will produce completely useless code, knowingly (because up to that phrase the reasoning was correct) breaking things.

Today another thing started happening which are phrases like "I've been burning too many tokens" or "this has taken too many turns". Which ironically takes more tokens of custom instructions to override.

Also claude itself is partially down right now (Arp 6, 6pm CEST): https://status.claude.com/