https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119414
Robert Dubner <rdubner at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|REOPENED |RESOLVED
Resolution|--- |FIXED
--- Comment #16 from Robert Dubner <rdubner at gcc dot gnu.org> ---
Mainframe programmers sometimes think differently, I believe.
If my understanding from discussions with an old-school mainframe programmer --
who admittedly hasn't coded in many years -- are correct, they happily do
things like
CALL "get-power-from-windmill"
ON EXCEPTION
CALL "get-power-from-waterwheel"
END-CALL
END-CALL
This lets you use the same code for different clients, one that has windmills,
and another that doesn't. You just specify a different list of procedures,
which they do routinely with the incomprehensible lists of Job Control Language
statements that abound in the mainframe world.
I have known Linux programmers to look at this and their brains explode. The
usual Linux ethic is that a missing global is an error, full stop, no
discussion.
In the old-school mainframe world, not finding a procedure can be normal. You
just do something else.
Okay. I am getting the idea. I will reclose this PR; I will use explicit gcc
options, including -rdynamic and -rpath and whatnot as necessary, and I will
investigate __attribute__((weak)).
I wonder why *I* didn't think of that. Oh, wait. I've never heard of it
before. Maybe that's why.
