On Aug 6, 2010, at 11:13 AM, Douglas Gregor wrote: > > On Aug 6, 2010, at 11:47 AM, Argyrios Kyrtzidis wrote: > >> Author: akirtzidis >> Date: Fri Aug 6 04:47:24 2010 >> New Revision: 110436 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=110436&view=rev >> Log: >> Introduce a new token kind 'cxx_defaultarg_end' to mark the end of C++ >> default arguments that were part of >> lexed method declarations. >> >> This avoid interference with tokens coming after the point where the default >> arg tokens were 'injected', e.g. for >> >> typedef struct Inst { >> void m(int x=0); >> } *InstPtr; >> >> when parsing '0' the next token would be '*' and things would be messed up. > > Good idea. > >> Modified: >> cfe/trunk/include/clang/Basic/TokenKinds.def >> cfe/trunk/lib/Parse/ParseCXXInlineMethods.cpp >> cfe/trunk/lib/Parse/ParseDecl.cpp >> cfe/trunk/test/Parser/cxx-default-args.cpp >> >> Modified: cfe/trunk/include/clang/Basic/TokenKinds.def >> URL: >> http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/TokenKinds.def?rev=110436&r1=110435&r2=110436&view=diff >> ============================================================================== >> --- cfe/trunk/include/clang/Basic/TokenKinds.def (original) >> +++ cfe/trunk/include/clang/Basic/TokenKinds.def Fri Aug 6 04:47:24 2010 >> @@ -96,6 +96,7 @@ >> TOK(eof) // End of file. >> TOK(eom) // End of macro (end of line inside a macro). >> TOK(code_completion) // Code completion marker >> +TOK(cxx_defaultarg_end) // C++ default argument end marker >> >> // C99 6.4.9: Comments. >> TOK(comment) // Comment (only in -E -C[C] mode) >> >> Modified: cfe/trunk/lib/Parse/ParseCXXInlineMethods.cpp >> URL: >> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Parse/ParseCXXInlineMethods.cpp?rev=110436&r1=110435&r2=110436&view=diff >> ============================================================================== >> --- cfe/trunk/lib/Parse/ParseCXXInlineMethods.cpp (original) >> +++ cfe/trunk/lib/Parse/ParseCXXInlineMethods.cpp Fri Aug 6 04:47:24 2010 >> @@ -142,9 +142,13 @@ >> OwningExprResult DefArgResult(ParseAssignmentExpression()); >> if (DefArgResult.isInvalid()) >> Actions.ActOnParamDefaultArgumentError(LM.DefaultArgs[I].Param); >> - else >> + else { >> + assert(Tok.is(tok::cxx_defaultarg_end) && >> + "We didn't parse the whole default arg!"); >> + ConsumeToken(); // Consume tok::cxx_defaultarg_end. > > How about we SkipUntil the tok::cxx_defaultarg_end, and give an error if it's > not the next token? That way, if expression parsing doesn't use up all of the > tokens, we'll give a reasonable diagnostic rather than asserting.
But if there was no error in expression parsing, clang should have used up all of the tokens, otherwise it's a bug; can DefArgResult be valid after expression parsing aborted due to an error ? What kind of diagnostic do you have in mind, wouldn't it seem strange since the user's code would have no actual problem ? -Argiris _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
