Extend HIR to track the source and syntax of a lifetime #139945
Conversation
An upcoming lint will want to be able to know if a lifetime is hidden (e.g. `&u8`, `ContainsLifetime`) or anonymous: (e.g. `&'_ u8`, `ContainsLifetime<'_>`). It will also want to know if the lifetime is related to a reference (`&u8`) or a path (`ContainsLifetime`).
rustbot
added
S-waiting-on-review
|
|
||
| /// Details not yet needed. Feel free to give useful | ||
| /// categorization to these usages. | ||
| Other, |
There was a problem hiding this comment.
What is this for? Can you give an example?
There was a problem hiding this comment.
The comment is attempting to express this, but it's literally just every other usage that isn't one of the previous categories. Those usages are not relevant to the lint I'm writing, and I'm not familiar with what they are, so it didn't seem right to spend time to categorize them in ways that are not useful to me and potentially actively un-useful for the next person.
There was a problem hiding this comment.
Sorry, I'm fine with having Other, but can you give one or two examples in the comment? It's not obvious to me what other cases there are.
| fn from(ident: Ident) -> Self { | ||
| let name = ident.name; | ||
|
|
||
| if name == kw::Empty { |
There was a problem hiding this comment.
Can kw::Empty ever reach here? If so, where from? I'm curious because I've been eliminating kw::Empty uses aggressively lately.
There was a problem hiding this comment.
I don't actually know; this was more of a "just in case" scenario. Do you want me to change this to a panic?
There was a problem hiding this comment.
If it doesn't trip on any test, yes please.
| } | ||
|
|
||
| #[derive(Debug, Copy, Clone, PartialEq, Eq, HashStable_Generic)] | ||
| pub enum LifetimeSyntax { |
There was a problem hiding this comment.
Are all combinations of LifetimeSource/LifetimeSyntax possible?
There was a problem hiding this comment.
No, I don't believe so. For example, I don't think you can have LifetimeSyntax::Hidden with LifetimeSource::OutlivesBound or LifetimeSource::PreciseCapturing
There was a problem hiding this comment.
That would be good to capture in a comment. Maybe on the definition of Lifetime, because that's where LifetimeSyntax and LifetimeSource are used together.
| }; | ||
| let span = self.tcx.sess.source_map().start_point(t.span).shrink_to_hi(); | ||
| let region = Lifetime { ident: Ident::new(kw::UnderscoreLifetime, span), id }; | ||
| (region, LifetimeSyntax::Hidden) |
There was a problem hiding this comment.
Here the name is '_ but the LifetimeSyntax is Hidden... why is that? It's the only call to lower_lifetime where the given ident doesn't match the ident in the lifetime.
There was a problem hiding this comment.
It has to have a name — ident isn't an Option and you are removing these kinds of usages of kw::Empty, so UnderscoreLifetime would be the next most semantic choice. This corresponds to &Type — there is no original source code name for the lifetime, it is hidden.
It's the only call to
lower_lifetimewhere the given ident doesn't match the ident in the lifetime.
That's true, but notably lower_lifetime_hidden_in_path and elided_dyn_bound do the same concept of pairing UnderscoreLifetime with Hidden.
|
☔ The latest upstream changes (presumably #139938) made this pull request unmergeable. Please resolve the merge conflicts. |
|
Adding the panic on the empty case and a couple more comments should be enough for this to merge, thanks. |
An upcoming lint will want to be able to know if a lifetime is hidden (e.g.
&u8,ContainsLifetime) or anonymous: (e.g.&'_ u8,ContainsLifetime<'_>). It will also want to know if the lifetime is related to a reference (&u8) or a path (ContainsLifetime).r? @nnethercote