Carl Lowenstein wrote:
Well, then let me object to the lack of warning.
Let me object to your depending on behavior that is a *BUG*.
Allowing your SCCS/RCS string through was an artifact of never
optimizing away unused static strings--that was a bug. Suddenly they
made the compiler do the correct thing, and now they have to put up with
complaints.
Man, I'd be grouchy, too.
By the way, reading the "gcc at gcc dot gnu dot org " mailing list I
see that I am not the only one who has used embedded strings for
tagging compiled files so they can be traced back to their sources, or
for other identification purposes.
Okay, lots of people depend on buggy behavior. Your point?
More work for me means that I have to remember __attribute__((unused))
to defeat this particular "optimization". It also means that I spent
an hour or two wondering why things that used to work no longer do.
Uh, noooo ... Your *program* worked fine.
What changed was that you couldn't grovel around the undocumented,
unspecified innards of the object files the way you used to.
Not feeling much sympathy here ...
Charge it up to education.
Indeed.
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg
