frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

1•spilledcoffee•11s ago

Everyone Is Wrong About AI Except Me

https://jasonpargin.substack.com/p/everyone-is-wrong-about-ai-except
1•teddyh•38s ago•0 comments

Managed to get a criminal investigation into adtech

https://www.thatprivacyguy.com/blog/criminal-investigation-adtech-surveillance/
1•AlexanderHanff•44s ago•1 comments

Ask HN: How is the personal assistant you're building different?

1•velyan•2m ago•0 comments

A friendly, non-technical introduction to differential privacy

https://desfontain.es/blog/friendly-intro-to-differential-privacy.html
1•simonebrunozzi•2m ago•0 comments

Show HN: Trial library based with OTP detected tempmail service

https://huskmail.xyz
1•draggodra•3m ago•0 comments

Today's Office: A Visual Log

https://www.ssp.sh/brain/todays-office/
1•zazuke•4m ago•0 comments

Selling Smart Link – Affiliate Tracking SaaS ($2k)

https://www.by-smartlink.com/
1•smartlink30•5m ago•0 comments

Europeans should learn to love the air-conditioner … green electricity

https://www.economist.com/europe/2026/06/18/europeans-should-learn-to-love-the-air-conditioner
1•Kaibeezy•5m ago•1 comments

No Side to Take: Political Indifference Inside the 4vps.su Leak

https://disclosing.observer/2026/06/23/political-indifference-4vps-leak.html
1•0x5h4un1•6m ago•0 comments

Talk of a bubble is 'blasphemy against AI' says SoftBank's Son

https://www.reuters.com/business/media-telecom/talk-bubble-is-an-insult-ai-says-softbanks-son-202...
3•1vuio0pswjnm7•8m ago•1 comments

Qualcomm to Acquire Modular

https://www.modular.com/blog/qualcomm-to-acquire-modular
1•colesantiago•8m ago•0 comments

Qualcomm to acquire Modular

https://www.qualcomm.com/news/releases/2026/06/qualcomm-to-acquire-modular
1•visheshdembla•8m ago•0 comments

ZerfAI – AI Micro SaaS Ideas from Reddit, HN and GitHub Complaints

https://www.zerfai.com/
1•awsaqh•8m ago•0 comments

Show HN: Xcreener – I built a simpler screener for forex traders

https://xcreener.com
1•guzzul•9m ago•0 comments

Show HN: Topaz – replace three Azure emulators with one binary

https://github.com/TheCloudTheory/Topaz
1•kamilmrzyglod•9m ago•1 comments

Uptime Monitor Shows green but visitors are getting errors

https://medium.com/@thesuperrepemail/your-uptime-monitor-shows-green-but-visitors-are-getting-err...
1•mssblogs•11m ago•0 comments

The AI Revolution: The Road to Superintelligence (2015)

https://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
1•cool-pear•12m ago•0 comments

Is it more unusual to have written a book, or to have killed someone?

https://homosabiens.substack.com/p/is-it-more-unusual-to-have-written
1•surprisetalk•12m ago•0 comments

The tech industry wants to take away your screen

https://www.bbc.com/future/article/20260624-why-tech-companies-want-to-take-away-your-screen
1•bookofjoe•13m ago•1 comments

Should data centers pay a carbon tax?

https://www.slowboring.com/p/should-data-centers-pay-a-carbon
1•thm•14m ago•0 comments

Simplicity Police

https://valeriavg.dev/simplicity-police
1•mooreds•15m ago•0 comments

Archaeologists find Viking textile production site in Denmark

https://text.npr.org/g-s1-129750
1•mooreds•16m ago•0 comments

Generating Images with a 2020 iPhone

https://www.duration.ai/blog/generating-images-with-a-2020-iphone
2•sudb•16m ago•0 comments

Slutcon: a three day sexuality education retreat.

https://www.slutcon.com/
1•MrBuddyCasino•17m ago•0 comments

The Drone That Has to Push Air to Stand Still

https://atomsfrontier.substack.com/p/the-drone-that-has-to-push-air-to
1•jpatel3•17m ago•1 comments

Codex Seraphinianus

https://en.wikipedia.org/wiki/Codex_Seraphinianus
1•josh-sematic•18m ago•0 comments

Juncheng Vehicle Co. is replicating iconic car bodies [video]

https://www.youtube.com/watch?v=cBBZrjwqWZc
1•msephton•20m ago•0 comments

Minimus container images are now free

https://images.minimus.io/
6•dimastopel•20m ago•1 comments

Show HN: Open-source language for analyzing, transforming, and modeling data

https://www.malloydata.dev/
1•nih567•24m ago•0 comments
Open in hackernews

Ask HN: How to avoid LLMs struggling with Lisp parens?

2•chriswarbo•1h ago
LLMs seem to love certain languages (Python, Bash, etc.), but they all seem to struggle with Lisp (e.g. Racket or Emacs Lisp). I've tried various iterations of Claude, as well as cheaper models like DeepSeekV4, etc. and the pattern is the same: they'll make a few successful edits, but eventually they'll get some parentheses slightly wrong, then spiral into madness as they attempt to fix the syntax errors by counting and matching-up parentheses "manually" in a never-ending loop.

This is frustrating for two reasons:

Firstly, LLMs are famously bad at counting characters (e.g. the number of "r"s in "strawberry"), so it's no wonder this approach of generating and counting characters doesn't work very well.

Secondly, balancing parentheses is trivial for traditional, non-LLM algorithms; so it feels like an entirely avoidable problem (without resorting to larger, more-expensive models).

Is anyone using LLMs successfully on Lispy projects? If so, what workflows, tooling, etc. have you found to work well? I've tried guiding them to use Emacs `check-parens` rather than counting "manually"; but maybe inferring from indentation might work better? Perhaps tree-based generation/tools would avoid introducing such problems in the first place?

Comments

backrun•1h ago
I suspect the most the reliable approach is to stop treating the model as the syntax checker. Let it propose a small diff, then run the result through check-parens, the language parser, or a formatter before accepting the change. if valifation fails, feed back the excat parser error and smallest affected form rather than the entire file