currently we use a ton of #[should_panic] attributes on our tests
most of them don't include an expect inside them, but some do. tests/bytes.rs:84:
#[should_panic(expected = "advance out of bounds: the len is 8 but advancing by 12")]
fn // ...
to make it consistent all #[should_panic] attributes on tests/*.rs should have an expect reason added to them
some of the exact errors are not deterministic, but the error type is, and since expect matches a substring it's easier to maintain if we just keep the error name without details
good example: "advance out of bounds"
bad examples:
- the full error with details: "advance out of bounds: the len is 8 but advancing by 12"
- no expect, just bare
#[should_panic]
- anything that does not match the error name
currently we use a ton of
#[should_panic]attributes on our testsmost of them don't include an expect inside them, but some do.
tests/bytes.rs:84:to make it consistent all
#[should_panic]attributes ontests/*.rsshould have an expect reason added to themsome of the exact errors are not deterministic, but the error type is, and since expect matches a substring it's easier to maintain if we just keep the error name without details
good example: "advance out of bounds"
bad examples:
#[should_panic]