A new C++ bug

A new C++ bug

[H.J. Lu]

Hi,

With egcs in CVS at 1:30pm PDT on May 06, 1988:

# gcc -B/home/work/gnu/bin/egcs/gcc/ -S x.cc
x.cc:24: Internal compiler error 378.
x.cc:24: Please submit a full bug report to `egcs-bugs@cygnus.com'.

It seems that egcs in CVS mangles both function names to
set_overflow_handler__FT0T1.


-- 
H.J. Lu (hjl@gnu.org)
---x.cc--
typedef          int  _G_int32_t __attribute__((__mode__(__SI__)));
typedef unsigned int _G_uint32_t __attribute__((__mode__(__SI__)));

typedef struct
{
  _G_int32_t            u;
  _G_uint32_t           l;
} twolongs;

typedef void (*Fix16_peh)(short&);
typedef void (*Fix32_peh)(_G_int32_t&);
typedef void (*Fix24_peh)(_G_int32_t&);
typedef void (*Fix48_peh)(twolongs&);
extern Fix16_peh set_Fix16_overflow_handler(Fix16_peh);
extern Fix32_peh set_Fix32_overflow_handler(Fix32_peh);
extern Fix24_peh set_Fix24_overflow_handler(Fix24_peh);
extern Fix48_peh set_Fix48_overflow_handler(Fix48_peh);

void
set_overflow_handler (Fix16_peh handler16, Fix32_peh handler32)
{
}

extern void set_overflow_handler(Fix24_peh, Fix48_peh);
void
set_overflow_handler (Fix24_peh handler24, Fix48_peh handler48)
{
}

Re: A new C++ bug

[Martin von Loewis]

> With egcs in CVS at 1:30pm PDT on May 06, 1988:
> 
> # gcc -B/home/work/gnu/bin/egcs/gcc/ -S x.cc
> x.cc:24: Internal compiler error 378.
> x.cc:24: Please submit a full bug report to `egcs-bugs@cygnus.com'.
> 
> It seems that egcs in CVS mangles both function names to
> set_overflow_handler__FT0T1.

That would cause the assertion to fail. It asserts that two functions
with the same assembler name really should be identical functions.

It seems that name mangling in the presence of typedefs doesn't work,
currently.

Regards,
Martin