Conversation

Apr 27, 2020

Friends learning Rust: writing about lifetimes is something I get requests for a lot, but it's hard for me to know exactly where people are stuck. If you're stuck, can you tell me where? Maybe you're reading the book and some part doesn't click? Ask about it with your words!

ᵉᵈᵈʸᵇ

@eddyb_r

Apr 27, 2020

I've had a hunch for ages and even ranted about it once in a while, years ago every time we teach or talk about named lifetimes as if they're not *generic parameters*, we make muddle the waters like pretending <'a> is coincidental syntax reuse and not the same as <T>

ᵉᵈᵈʸᵇ

@eddyb_r

Apr 27, 2020

another lie is "concrete lifetime". there really isn't such a thing, especially not with NLL other than I guess 'static everything is parameters and threading relationships through data structures and function signatures

Brendan Zabarauskas

@brendanzab

Apr 28, 2020

Existential lifetimes when? Super frustrating when NLL cna figure out a thing I write, but then I can't split it out into a separate function/definition 😖

Replying to

and

To clarify, NLL is amazing and awesome and I'm glad it's in there! But yeah, it does mean you can run into situations where you literally can't split something into smaller parts. 😅

4:11 AM · Apr 28, 2020Twitter Web App

Replying to

and

that's... intentional. you'd need much more complexity in what you can describe in signatures to be able to handle all of the intra-function situations borrowing only a subset of fields, or even with separate borrows, is the biggest thing AFAIK

Brendan Zabarauskas

@brendanzab

Apr 28, 2020

I guess one thing I wish for is `fn foo() -> exists<'a> Bar<'a>`, which can be pretty handy for testing. But yeah, I guess in the general case it could definitely get mighty complicated…