C++ demangler horrors

[David Carlton]

On Fri, 27 Jun 2003 15:00:26 -0700, "H. J. Lu" <hjl@lucon.org> said:

> binutils and gdb want to use a C++ demangler written in C++ with a C
> interface. That is the demangler needs a C++ compiler to compile,
> but not libstdc++ to use.

Maybe I've missed something, but I haven't seen any discussion on the
GDB mailing list that would make me think that requiring a C++
compiler to compile GDB would be acceptable right now.  It would be
great if it were acceptable, but if that's what you are proposing, I
think that it should be raised on gdb-patches.

I'm also dubious about having GDB use one of two different demanglers
depending on whether or not a C++ compiler is present.  Unfortunately,
some current changes to GDB are, at least in the short term, making it
more dependent on the exact output of the demangler, not less
dependent.  (Then again, the output of the current demangler isn't
always ideal for our purposes.)

David Carlton
carlton@kealia.com