phiresky

Might be a good idea to separate the postgresql stuff (adapter and type definitions) into a separate package like `@hediet/typed-sql-postgres`. or `@typed-sql/postgres` Not sure how the serializer / etc can...

That's a good idea - but I think backprop is pretty hard to visualize. I recommend you watch [the video on backpropagation by 3Blue1Brown](https://www.youtube.com/watch?v=Ilg3gGewQ5U) (if you haven't seen it yet)...

> Looks like you also support some types of persistent identifiers directly in the citekey Yep, but only doi: and isbn:. I think it's usually better to use URLs anyways...

Actually I just found the manubot.org homepage which explains stuff better - my fault haha. I just went through the repos, maybe link from the rootstock repo to the homepage...

> And this would define the @abc citekey for use everywhere? That seems like a nice syntax. Yeah. Imo it's kind of a mistake that pandoc decided to introduce a...

> Or did I misunderstand something and the filter is not intended to be used in combination with normal citekeys? Right now there's four ways you can use url2cite in...

It looks like you have some parsing error. Please send the content of your document (/ a minimal reproduction)

> I would actually suggest to maybe make url2cite-allow-dangling-citations=true the default option That might be true.. though this would make it harder to spot errors for the "ideal" case, and...

A possible solution would be to be able to import the `messages.html` from facebook export..

I think they now also allow JSON export, which is even easier to parse. Only problem is the file size so it probably needs a streaming JSON parser. Shouldn't be...