I'm pleased to announce a #hledger double release! First, hledger 1.52, with
more robust paging,
faster valuation,
commodity tags,
and cost basis annotations which are preserved and can be exported (examples provided).
- https://github.com/simonmichael/hledger/releases/1.52
- https://hledger.org/relnotes.html#2026-03-20-hledger-152
- https://hledger.org/install
And, hledger 2.0 preview 1 (AKA 1.99.1), which adds
automated lot tracking and capital gains calculation,
improvements to `print`'s beancount output,
and a new policy for AI-assisted development.
This is the first hledger release to explore AI-assisted development, so please do check out the evolving AI policy/FAQ: https://hledger.org/AI.html
- https://github.com/simonmichael/hledger/releases/1.99.1
- https://hledger.org/relnotes.html#2026-03-28-hledger-1991
Jensen Huang approves of this message.
> there’s much grey area between clear plagiarism/theft, and fair use/independent invention.
Say that again to the gitforges that are swamped with ai scraping.
> optimise, and limit our AI resource usage (measured by tokens in & out, eg).
So no matter how much resources was required to make the llm in the first place.
> It will continue to become more efficient and less costly in resources
Have you heard about the Jevon’s paradox? Every new “advancement” in ai requires more tokens, and every new ai techbro requires more datacenter ram/gpu’s
> once you have experienced the labour saving of a good AI setup
“Labour saving” is bourgeoisie speak for lowering wages and mass firing.
@simonmic I enjoyed your well put together statement here Simon: https://hledger.org/AI.html
This is a great, transparent approach to adopting these tools responsibly and with care. Thank you!
Simon Michael
lil5
Atlas
Caleb Maclennan