Working ELF GCC

Nick Burrett nick at dsvr.net
Tue Oct 7 05:21:11 PDT 2003


Peter Naulls wrote:
> I also recall a discussion on the FORTRAN
> compiler that there wasn't a situation AOF wasn't able to cope with.

Ah yes.  The FORTRAN problems consists of a situation like this:

In one object file:

    AREA  |C$$data|
|foobar|
    DCB 0,1,2,3,4,5,6


In another object file
    AREA  |foobar|, COMMON
    % 7

i.e. we create a COMMON block that is initialised with values in the 
first object file.  The second object file merely makes reference to 
that COMMON block.

I couldn't figure out an answer at the time.  It might be fixed by the 
changes I recently made to support the C++ library.

However I decided to drop support for the F77 compiler because it wasn't 
up to a standard that the Acorn Fortran users thought they could live 
with.  The compiler code was too difficult to follow and understand and 
the whole lot was declared dead by the original author.  Fortran will 
live again through a port of G95 which is destined for GCC 3.5.  This is 
going a bit off-topic so I'll stop there.


-- 
Nick Burrett
Network Engineer, Designer Servers Ltd.   http://www.dsvr.co.uk





More information about the gcc mailing list