This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: A completely different approach to EH runtime


>>>>> "Geoff" == Geoff Keating <geoffk@geoffk.org> writes:

    Geoff> I didn't think that you could use the same compiler on both
    Geoff> Red Hat 6.x and Red Hat 7, because the libio was different.
    Geoff> Certainly here at Red Hat's engineering section we have
    Geoff> different compilers for x86-linux-gnu, x86-linux-gnulibc2.1
    Geoff> and x86-linux-gnulibc2.2.

In my experience, I can use the same gcc-2.95.2 executable
successfully on Red Hat 6.0 and 7.0.  It might help that the gcc
exectuable was built on the older system, but if so, it is just the
least-common-denominator effect w.r.t. glibc, not a gcc thing per se.
It may also help that I mostly do number crunching and simple file
I/O.  No threads, no graphics.  Recently I've tried using C++
exceptions and had some strange problems, which I cannot reproduce,
but which seem to be related to getting the right exception code into
things. 

    Geoff> Between libio, fixincludes, libgcc, thread libraries, and
    Geoff> ptrace, I think that similar things are true for Linux,
    Geoff> Solaris, and AIX: if you upgrade the OS, you need to
    Geoff> rebuild the tools.

In my experience with SunOS, IRIX, and HPUX, I generally have re-built
gcc for major new releases of the OS, but it is not necessary to
re-build for minor release numbers.  Thus in our exec-prefix,
/tsri/gnu/${archosv}, we have sun4SunOS4 and sun4SunOS5 as possible
values for ${archosv}, but not sun4SunOS5.2, etc.

-Don


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]