Rework ForEachStmt Desugaring #86010
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
elsakeirouz
commented
Dec 12, 2025
elsakeirouz
force-pushed
the
rework-for-each-desugar
branch
2 times, most recently
from
5381915 to
e7ec1ab
Compare
elsakeirouz
commented
Dec 13, 2025
elsakeirouz
commented
Dec 13, 2025
elsakeirouz
commented
Dec 13, 2025
| return ctx.getAsyncIteratorNext(); | ||
| } | ||
|
|
||
| static BraceStmt *desugarForEachStmt(ForEachStmt* stmt){ |
Author
There was a problem hiding this comment.
Potentially refactor this function to make it more readable.
elsakeirouz
force-pushed
the
rework-for-each-desugar
branch
from
0dc0a64 to
714b32f
Compare
I have not yet tested any behavior of generated SIL code but it seems reasonable in comparison. I was able to compile the standard library. Here: * include/swift/AST/ASTBridging.h, * include/swift/AST/Expr.h, * include/swift/AST/ExprNodes.def, * include/swift/AST/Stmt.h, * include/swift/AST/StmtNodes.def, * include/swift/AST/TypeCheckRequests.h, * include/swift/AST/TypeCheckerTypeIDZone.def, * include/swift/Sema/ConstraintLocator.h, * include/swift/Sema/SyntacticElementTarget.h, * lib/AST/ASTDumper.cpp, * lib/AST/ASTPrinter.cpp, * lib/AST/ASTScopeCreation.cpp, * lib/AST/ASTVerifier.cpp, * lib/AST/ASTWalker.cpp, * lib/AST/Bridging/StmtBridging.cpp, * lib/AST/Expr.cpp, * lib/AST/Stmt.cpp, * lib/AST/TypeCheckRequests.cpp, * lib/ASTGen/Sources/ASTGen/Stmts.swift, * lib/Parse/ParseStmt.cpp, * lib/SILGen/ASTVisitor.h, * lib/SILGen/SILGenExpr.cpp, * lib/SILGen/SILGenStmt.cpp, * lib/SILOptimizer/Mandatory/MoveOnlyDiagnostics.cpp, * lib/Sema/BuilderTransform.cpp, * lib/Sema/CSApply.cpp, * lib/Sema/CSDiagnostics.cpp, * lib/Sema/CSGen.cpp, * lib/Sema/CSSimplify.cpp, * lib/Sema/CSSyntacticElement.cpp, * lib/Sema/SyntacticElementTarget.cpp, * lib/Sema/TypeCheckEffects.cpp, * lib/Sema/TypeCheckStmt.cpp.
elsakeirouz
force-pushed
the
rework-for-each-desugar
branch
from
714b32f to
5285f67
Compare
elsakeirouz
force-pushed
the
rework-for-each-desugar
branch
from
67a017d to
b5e88da
Compare
elsakeirouz
force-pushed
the
rework-for-each-desugar
branch
from
a985680 to
8dbec9f
Compare
elsakeirouz
force-pushed
the
rework-for-each-desugar
branch
from
bd1a68e to
623bdc5
Compare
Comment on lines
+4201
to
+4202
| // FIXME: need to find out if the stmt is part of a foreach's | ||
| // desugared |
Contributor
There was a problem hiding this comment.
Isn't this issue resolved by the opaque pattern fix?
Contributor
There was a problem hiding this comment.
@elsakeirouz I think the issue is that markOwnedByStatement is walking into the opaque pattern, I think forEachVariable should probably have a parameter that controls whether it should walk into them
Contributor
There was a problem hiding this comment.
It might also be worth making it so we don't walk into opaque patterns by default in forEachVariable, only SILGen and maybe a couple of Sema clients should care about that case
Contributor
There was a problem hiding this comment.
I think the same also applies to getSemanticsProvidingPattern, I think we should see if we can get away with not recursing into opaque patterns by default
Co-authored-by: Hamish Knight <[email protected]>
Co-authored-by: Hamish Knight <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR aims to lift the desugaring of ForEachStmt (when dealing with Sequences) from SILGen to the AST, and simultaneously delaying the synthesis of makeIterator and next calls until after typechecking.
The motivation for this is support for non-Copyable & non-Escapable sequences with non-Copyable elements in ForEachStmt. In order to know which variant of makeIterator and next calls we need, we must determine whether a Sequence is copyable, asynchronous, or neither. The compiler currently relies on the function choice to infer the type of the elements. However, to determine whether a Sequence is copyable, it must be typed.
For that reason, instead of matching the pattern type with the unwrapped result type of next(), we're now matching it with Sequence. This allows us to delay the decision and synthesis of those function calls until later.
Additionally, in order to avoid having to translate the rest of the complex desugar logic for the borrowing ForEachStmt in SILGen, we chose to represent it directly at the AST level, which SILGen will then use to emit the corresponding SIL code.
To make things simpler all around, we chose to start by reworking this pass for the regular/async ForEachStmts before handling borrowing statements.
We expect to be able to remove ConstraintKind::ValueWitness with this work.