kehvyn

If it helps, I've mostly been using AI to implement things in the craziest languages I can justify.

I write Typescript and SQL by day, my last two personal projects were Rust and Perl.

I do worry that I'm not learning them as deeply, but I am learning them and without AI as an accelerant I probably wouldn't be trying them at all.

I know exactly which store you're talking about, and I couldn't resist going in the one time I was in Vienna. It's a great store.

A fellow Shin Hanga-era fan! There are dozens of us!

Hasui is excellent, but Shiro Kasamatsu is my undisputed favorite. Something about the way he does buildings, and the dramatic colors, is exactly aimed at me.

It's always interesting to me that these plaintext sites are flagged as "insecure" and "risky" by modern browsers. I don't have a good solution, but it reminds me of [1]

[1](https://meyerweb.com/eric/thoughts/2018/08/07/securing-sites...)

Merry Christmas everyone! My friends and family are on two different continents so it's very quiet this year.

I hope everyone has a wonderful, relaxing, and joyous time.

A brief summary of a Postgres extension I enjoyed finding last year. I've used it a few times on side-projects.

I know it's not AS good as full-text search, but a lot of applications are still searching on specific fields with "WHERE name ILIKE "%John$", and this is a drop-in replacement.

I like the bigger touch targets in the header for mobile, but the infinite scroll hurts me, pages introduce natural stopping points and mean the header is always nearby.

(I appreciate that it's Load More instead of automatic infinite scroll, but still!)

Oh that YouTube URL feature, that's great. You predicted my use case.

I've long wanted a pipeline to let the AI accumulate notes on the videos I watch, this is perfect.

Fortunately the Met's data is public, both via API and a 330MB CSV. https://github.com/metmuseum/openaccess

Essentially they have an "Object Date" field that's a human-readable string and could be anything, and then they include "Object Start Date" and "Object End Date" that are integer years so that it's machine readable and you can do those comparisons.

I actually looked at supporting dates like this, but if you go through the Met's open dataset (https://github.com/metmuseum/openaccess) that kind of "alternative calendar with no reference to the BCE/CE dates" is basically nonexistent.

There are references to the Islamic and Japanese calendar systems, but always next to the CE equivalent.

Data entry is fortunately being done by modern people, so the translation to CE/BCE is usually baked in, and all you need to support is every possible way somebody could say "the early half of" and "5th millenium B.C. to mid 1914"

I've actually done this, and it's very fun.

My main testing dataset is the 470,000 records from the Met, with 33k unique date values. Fortunately they include machine-readable dates I can validate against.

https://github.com/kjrocker/epochal

I think the anticipation of interruption caused by meetings might be worse, I end up losing a half hour on both ends.

The USA version is just the letters ADA in gigantic font.

Europe still hasn't caught up to ADA. I don't know any other really good laws that are unique to the US, but I'm sure they exist.

Wow, that's a lot of AI.

Do you think they're all using actual LLMs? I've got a natural language parser I could probably market as "AI Semantic Detection" even though it's all regular expressions