================ Comment at: include/clang/AST/Stmt.h:1962 @@ +1961,3 @@ + /// \brief The implicit outlined function. + FunctionDecl *TheFuncDecl; + ---------------- Wei Pan wrote: > Doug Gregor wrote: > > Interesting. We're actually going to synthesize a FunctionDecl for captured > > statements? I would have expected that the function would be synthesized > > entirely in CodeGen, and would never be represented in the AST. Why do we > > need to represent the function in the AST? > This FunctionDecl was synthesized during Sema and we marked this function as > a capturing region. With this approach, its codegen seems simpler, comparing > to the implementation for "CodeGenFunction::GenerateCopyHelperFunction(const > CGBlockInfo &blockInfo)" > > What about we first post our parsing and sema patches? If having this > FunctionDecl is not ideal, we would certainly like to remove it out of the > AST and generate it during CodeGen. Seemly, we do need this FunctionDecl as the DeclContext for anything inside this captured statement, just like Lambda::operator()() or BlockDecl.
http://llvm-reviews.chandlerc.com/D370 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
