Conversation

I still want to have my cake and eat it with regards to leveraging abstractions from mathematics, but also keeping things learnable for humans, and sympathetic to the real machines we run things on. Also see my comment here: users.rust-lang.org/t/cgmath-looki
Quote Tweet
Replying to @brendanzab and @ozkriff
All I know is that it's a hard problem that I bounced back and forth on over the years and never ended up satisfied. I'm beginning to think that languages themselves need better support for helping users to move from the concrete to the abstract as they learn a domain.
Replying to
Kind of feel like I want to be able to instantiate away certain abstractions to get to the concrete stuff early on, but then remove that as I get more comfortable. This should be able to happen in all aspects of the language tooling (IDE, docs, error reporting, etc...)
1
It's great to be at an Elm level of complexity early on, but I don't want to trap people there. I also want to help to guide library designers to build things that avoid weird special-cases that are hard to abstract over. This is hard if APIs are stuck being too concrete.