Error handling August 20, 2024 on Runxi Yu’s Website

I haven’t found a single language with error handling to my liking:

• C: errno sucks and I haven’t seen anyone that disagrees yet, and the thing some libraries like OpenSSL does with its custom error type and detecting the underlying errno is even more annyoing
• Hare (and to some extent, Rust, though I haven’t gotten to really know Rust yet): IDK, it just feels really unnatural to me (I’ll see how I could explain in the future)
• Go (and to some extent, Linux x86 asm): Hella boilerplate (especially in Go without macros)
• Python, and most other VM’ed languages that use exceptions: Functions in between the throw’er and the catch’er don’t get a say in what’s going on, which also means as a catcher you don’t know where that ValueError came from (was it in integer conversion or decoding or what?)
• Haskell: Kinda a mess, see Hare/Rust for Maybe/Either; see Python for exceptions; error can only be caught in IO monads which isn’t the majority of stuff in idiomatic Haskell
• POSIX shell: I’ve basically taken it to define a function a that runs its arguments and exits on failure (which is the most appropriate thing to do for simple scripts), but handling $? is still kinda hellish, and don’t even mention set -e

But at the same time sometimes I find exceptions to be very useful (e.g. say I’m handling a command in a network connection, and I want a catchall “command failed” that I could trigger by throwing an exception; I could emulate this with a goto up the stack but that also doesn’t feel right)