Still a lot of C++ files getting "fixed"

Mark Mitchell
Tue Jan 25 13:05:00 GMT 2000

>>>>> "Phil" == Phil Edwards <> writes:

    >>  True.  But, -ansi is specifically documented to be
    >> pedantically ANSI; that's its whole purpose in life.

    Phil> Very true.

    Phil> Since the C99 standard now specifies the same results as
    Phil> C++98, wouldn't "pedantically ANSI" now /require/ the new
    Phil> behavior? 

Well, that's a question about what ANSI means.  I think we're going to
have modes for "GNU C", pedantically ANSI/ISO C89, pedantically
ANSI/ISO C99, etc.  Probably, we'll have way more modes than we
should. :-) (One risk with open source software is that people tend to
add whatever features are convenient for them at the moment -- and
that can lead to a maintenance nightmare, an overly complex user
interface, etc.)

Kaveh's point that we can perhaps treat system headers differently
from user source is well-taken; I suppose it might make sense to
accept `//' comments only in system headers, under the assumption that
nowhere does there a lie a system header where this will change the
semantics of the of the program.

Mark Mitchell         
CodeSourcery, LLC     

More information about the Gcc-bugs mailing list