[CS] Invalidate nested unresolved VarDecls when ignoring completion argument #85142
Conversation
…rgument Make sure we set types for any nested VarDecls in UnresolvedPatternExprs to ensure we don't crash when attempting to solve the body.
hamishknight
requested review from
bnbarham,
hborla,
rintaro,
slavapestov and
xedin
as code owners
|
@swift-ci please test |
|
@swift-ci please SourceKit stress test |
| if (CS.isArgumentIgnoredForCodeCompletion(expr)) { | ||
| // Make sure we invalidate any nested VarDecls to ensure the body | ||
| // VarDecl for a case statement still gets a type assigned. | ||
| expr->forEachUnresolvedVariable([&](VarDecl *VD) { |
There was a problem hiding this comment.
I think a more robust long-term solution here might be to determine whether arguments are effectively ignored during pre-check and replace each one with something like a placeholder expression (something we'd need to add) that would generate a type variable or maybe remove them outright (I think we added some logic to the matcher at some point to not add "missing argument" fixes for cases like that).
There was a problem hiding this comment.
Unfortunately this is made trickier by the fact that we populate the case body variables when we create the AST, so we can't just drop AST nodes to fix this since the case body variables are still accessible to name lookup and won't have types assigned. Maybe we can come up with a better way of handling this
There was a problem hiding this comment.
We could do
expr->forEachUnresolvedVariable([&](VarDecl *VD) { ... };
as part of pre-check removing of the expressions...
2 participants
Make sure we set types for any nested VarDecls in UnresolvedPatternExprs to ensure we don't crash when attempting to solve the body.