[cfe-dev] parse C++ throw

[Mike Stump]

On Feb 24, 2008, at 8:24 PM, Chris Lattner wrote:
> Tabs:
>
> +  virtual ExprResult ActOnCXXThrow(SourceLocation OpLoc,
> +				   ExprTy *Op = 0) {

find . -name \*.h -o -name \*.cpp | xargs grep -c '      ' | grep -v ': 
0$'
./AST/Decl.cpp:46
./Driver/ASTConsumers.cpp:5
./Driver/clang.cpp:1
./include/clang/AST/ASTContext.h:1
./include/clang/AST/Decl.h:3
./include/clang/AST/DeclObjC.h:13
./include/clang/Parse/Action.h:3
./include/clang/Parse/Parser.h:2
./Parse/ParseDecl.cpp:1
./Parse/ParseExprCXX.cpp:3
./Sema/Sema.h:1
./Sema/SemaDecl.cpp:5
./Sema/SemaDeclObjC.cpp:1
./Sema/SemaExprObjC.cpp:4

:-(  Apparently I'm not the only one.  Do you think the number of  
people that miss set their physical tabs is larger than the number of  
people that use emacs?  Why not punish the people with wrong/weird  
tabs settings over punishing people that use emacs?  After all, we can  
merely educate people to use softtabstop and shiftwidth appropriately.

If not that...  could we add:
   /* -*- indent-tabs-mode:nil; -*- */
the tops of all the files or find some other way reduce the burden for  
emacs users?

Catering to people that think tabs in source files mean something  
other than 8, is like catering to people that like 6 bit characters or  
60 bit words, I'd rather just laugh at them.  Burdening all developers  
to `do something' doesn't strike me as friendly.