thicket

If you’re going to use SQLite regularly, you’ll need some extra infrastructure. OP’s post provides some boilerplate that would help you on your way.

What has really made the difference for me recently though, is Simon Willison’s sqlite_utils package. It’s heavier weight than OP’s wrapper classes, but it’s infinitely more powerful and has made using a local database a joy instead of the kludge I usually make of it. I’ll be bummed to writing stuff for PostGres when I have to do that

+1 for this question. Postman was a thing of simple beauty before they tried chasing the VC money

Even more concerning is all the upvotes. That means somebody’s not only posting bullshit, they’re running bots here too

This is a hero list! Props

Nice to hear. So… anybody using it yet? We’re all waiting to see if something’s better, cheaper, or easier than OpenAI.

Agreed. When I’ve had to make local carts every couple years, I have to spend 3 hours hacking a process together and remembering it all. I was hoping somebody would do something that “just works”, but this still seems like more conceptual burden than I was hoping for

Great question man. This is a big help to see!

So… what’s this article about? What did you find interesting about it?

Wow, the tone here is… really arrogant. When somebody starts off calling their audience idiots, I find it a little harder to read along.

I think I’m glad I did, though. Tone aside, I think it’s a worthwhile insight to note that caching is making up for a shortcoming in your data supply, and that fixing that shortcoming if at all possible should be a priority. The author’s summary would have helped me in the past:

Caching is a useful tool, but can be easily abused without giving any signs of the abuse.

Don’t get involved with caching till the last minute; find any other way you can first. Optimise your application before you use the blunt tool of caching.

While I would love to see Twitter auguring straight into the ground, Twitter’s API changes would explain some portion of this traffic changes. I wonder if there are any other proxy measures for audience engagement as separate from basic traffic