Conditional formats #362
Conditional formats #362
Conversation
|
Looks great! We're missing cases in |
Ahh thank you. Would be ideal if these could be compilation errors but a quick look suggests that's not super easy without lots of verbosity. |
|
Yeah, it's a bit frustrating that we don't get any errors there :( |
What do you think of this as a workaround for that? #[test]
fn value_has_unify_and_is_equal_impls() {
let value = Arc::new(Value::ConstLit(Const::Bool(false)));
// This test exists in order to cause a test failure when `Value` is changed. If this test
// has failed and you have added a new variant to Value it is a prompt to ensure that
// variant is handled in:
//
// - surface::elaboration::Context::unify
// - core::semantics::is_equal
//
// NOTE: Only update the match below when you've updated the above functions.
match value.as_ref() {
Value::Stuck(_, _) => {}
Value::Universe => {}
Value::FunType(_, _, _) => {}
Value::FunLit(_, _) => {}
Value::RecordType(_, _) => {}
Value::RecordLit(_, _) => {}
Value::ArrayLit(_) => {}
Value::FormatRecord(_, _) => {}
Value::FormatCond(_, _, _) => {}
Value::FormatOverlap(_, _) => {}
Value::ConstLit(_) => {}
}
} |
|
Oooh, I like that. Clever idea! |
doc/reference.md
Outdated
| The [representation](#format-representations) of a conditional format is the same | ||
| as the format without the condition when the condition evaluates to `true`. If | ||
| the condition is `false` then it's equivalent to the [fail format](#fail-format). |
There was a problem hiding this comment.
I'm not sure this is entirely accurate… this kind of implies something like Repr format | Void? Where as we want something more like:
| format | Repr format |
|---|---|
{ x <- format | cond } |
Repr format |
Not sure how to express this better in words, but it might be more accurate to say that “the representation of a conditional format is the same as the representation of the format guarded by the predicate”. It's a bit clunky to describe in words as we don't really have a name for what we're talking about. 😕
There was a problem hiding this comment.
“the representation of a conditional format is the same as the representation of the format guarded by the predicate”
Happy to change it to this. It's basically what I was was trying to say 😅
wezm
force-pushed
the
conditional-formats
branch
2 times, most recently
from
a0b0e98
to
9d9da01
Compare
Introduce support for conditional formats. Fixes #339
This allows values to be verified against a condition. E.g.
If the predicate succeeds then the format is read as usual. If it fails this currently results in an error being returned from
read_format, which in turn is presented as a panic/internal compiler error. I'm opening this PR to get the work reviewed so far and also to discuss how that fail case should be handled. I also need to update the reference and add some more tests when everything is locked down.