Re: name mangling in gcc 2.95.2

[pkurpis at keck dot hawaii dot edu (Peter Kurpis)]

Oops.  My mistake.

Turns out both compilers are aware of the  extern "C" { }
construct (enclosing C function prototypes to avoid the
mangling).

A more painstaking check revealed a header that was 
corrupted.

----- Begin Included Message -----

I get different name-mangling behavior between 
essentially vanilla installations of 2.91.66 and 
2.95.2.  

I have a C++ file making calls into a library that was 
written in C.  2.91.66 compiles the C++ file and does not 
mangle the C function call names; 2.95.2 mangles some of 
them -- making them incompatible with the entry points in 
the C library -- but I can't see the pattern.

I couldn't find anything in the documentation, except
 -fname-mangling-version-<n> , which implies there was
a change at 2.8, but not between 2.91 and 2.95.

Could someone help me, perhaps by pointing me at the right
doc?  I need to understand how g++ makes its decision 
whether to mangle or not to mangle -- that is the question.

Thanks in advance!




----- End Included Message -----