today I'm thinking about how floating point numbers can be treacherous -- what are specific examples of when they've betrayed you?

so far I have:

* money calculations where you need to get the exact correct answer (you have to use integers, not floats)
* twitter's tweet IDs are bigger than 2^53 and can't be represented as a Javascript number, you have to use a string

I'm looking for specific examples of what code you were writing that broke, not general facts about floats

this floating point post had so many great replies that I had to write a custom Python program to download them all so that I could really read them properly

(well, you could argue that I didn't technically *have* to, but programmers gotta program)

downloading mastodon replies to a post is surprisingly hard, I ended up just downloading all my mentions and statuses and then writing a recursive CTE in SQL to get all of the replies to a given post.

There's a `/context` endpoint that returns replies to a status but it doesn't return all of them and it doesn't seem to support pagination.

@b0rk Have you checked how all the replies are pulled when you view the post on Mastodon's web interface? Because that definitely shows them all.
@b0rk Hmm, it's hitting /context once. Are you sure that's missing some?
@b0rk Okay it's not actually unlimited, it's just set to such a high number that most of the time it returns all.
@b0rk Ah, but it's also capped at lower levels if it's an anonymous API request. Maybe that's the issue you were hitting?
@effy wow yes I think that was the issue, thank you so much!!
@b0rk @effy not sure if this was mentioned but if you’re looking at replies to someone else’s toot you want to ask their server. Your server will only know the ones from people that are followed by people on your server. Their server will have them all.