-
Notifications
You must be signed in to change notification settings - Fork 12.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Create type aliases for unresolved type symbols #45976
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Largely looks good, though I think this needs fourslash and possibly syntax server tests, specifically for
- find all references
- go to definition (at least the fact that it doesn't work)
- quick info
@@ -4520,7 +4523,7 @@ namespace ts { | |||
const noTruncation = compilerOptions.noErrorTruncation || flags & TypeFormatFlags.NoTruncation; | |||
const typeNode = nodeBuilder.typeToTypeNode(type, enclosingDeclaration, toNodeBuilderFlags(flags) | NodeBuilderFlags.IgnoreErrors | (noTruncation ? NodeBuilderFlags.NoTruncation : 0), writer); | |||
if (typeNode === undefined) return Debug.fail("should always get typenode"); | |||
const options = { removeComments: true }; | |||
const options = { removeComments: type !== unresolvedType }; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Would be helpful to leave a comment on this one (pun intended).
const options = { removeComments: type !== unresolvedType }; | |
// The unresolved type gets a synthesized comment on `any` | |
// to hint to users that it's not a plain `any`. | |
// Otherwise, we always strip comments out. | |
const options = { removeComments: type !== unresolvedType }; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sure, I'll add a comment.
src/compiler/checker.ts
Outdated
@@ -8255,6 +8269,10 @@ namespace ts { | |||
return type && (type.flags & TypeFlags.Any) !== 0; | |||
} | |||
|
|||
function isErrorType(type: Type) { | |||
return type === errorType || !!(type.flags & TypeFlags.Any && type.aliasSymbol); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why not just this? Seems like it'd be cheaper too.
return type === errorType || !!(type.flags & TypeFlags.Any && type.aliasSymbol); | |
return type === errorType || type === unresolvedType; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, that wouldn't work. I'll add a clarifying comment, but basically we want to identify the special TypeFlags.Any
types produced by getTypeForTypeAliasReference for an unresolved symbol. The sole purpose of unresolvedType
is to act as the declared type for the special type alias symbols we create for unresolved names. The type is never actually returned by anything because getTypeForTypeAliasReference maps it into a special TypeFlags.Any
type with an aliasSymbol
.
@@ -4,6 +4,6 @@ async function foo(): Promise<void> { | |||
|
|||
// Legal to use 'await' in a type context. | |||
var v: await; | |||
>v : any | |||
>v : await |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is there some way to make the type writer a little smarter here so that the team can differentiate between an unresolved
and a concrete type?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We certainly have the ability to tell the difference, but what did you have in mind in terms of output? Remember that we're often dealing with composed types, e.g. for var v: Promise<await[]>
we can't easily attach a comment to the await
identifier in the output.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A parenthesized type might work like (/*unresolved*/ await)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not really sure we want to do that. We're already generate an error at every reference to an unresolved symbol, I don't think we also want the extra noise in the type baselines.
I guess this doesn't technically fix #38836 because this only applies to types, right? |
Right, this doesn't fix #38836. One complication there would be what scope to introduce the unknown value symbols into. Should they be locals of the innermost enclosing function, or just globals like we do for the type symbols? The latter typically makes sense for types, but less clear for unresolved value symbols. |
I think just a global scope makes enough sense; if you do have a bunch of unresolved variables spread throughout your codebase, it's convenient to find-all-references, and likely to be a global anyhow; if it's mostly local (e.g. a user forgot to locally declare it), it doesn't really matter where you put it. But I'd understand if we wanted to take it one step at a time. |
Looks like there's a few conflicts that need to be resolved. |
# Conflicts: # src/compiler/checker.ts # tests/cases/fourslash/findReferencesJSXTagName.ts # tests/cases/fourslash/tsxFindAllReferences10.ts
With this PR we create type aliases for unresolved symbols such that quick info and error messages list back the unresolved name instead of just
any
.In this example where
ItemData
is not defined:hovering over
item
anditems
now showsItemData
andItemData[]
where previously we'd just showany
andany[]
. Furthermore, hovering over the unresolvedItemData
reference showstype ItemData = /*unresolved*/ any
.Fixes #45893.