Re: Dynamic size spec for widgets
On Thu, 24 Feb 2000 14:59:11 +0100, Hrvoje Niksic said:
Yeah, we discussed it in Japan. Linus acted a loser because his
code
bogusly _depended_ on inlining, meaning that it actually broke if the
functions were not inlined. We're not in the same situation here, not
even in the same ballpark.
OK.. I'll bite.. How *do* you write code that breaks if it's not inlined?
Only thing I can think of is if when not inlined on a multiprocessor,
hitting a race condition on updating a local variable that was optimized
away into a register when inlined?
--
Valdis Kletnieks
Operating Systems Analyst
Virginia Tech