Patch to execve

Bill Hacker wbh at conducive.org
Sat Feb 26 23:33:19 PST 2005


Kevin M. Kilbride wrote:

Kevin M. Kilbride wrote:

Secondly, it not only warns about _stripping_ qualifiers, but, in 
direct violation of the standard, it also warns about the 
explicitly-permitted conversion of an unqualified pointer to a 
qualified one.


Actually, in playing with this further, I can see that it is the base 
GCC compiler that emits this warning, not the -Wcast-qual option. This 
is much worse. The warning cannot be suppressed, even though it is 
warning about obviously permitted behavior.
I apreciate your thorough research, but do not see this as a 'show stopper'.

Imperfect, yes.  Inconsistent, yes.

In need of correction - 'probably', but not necessarily 'absolutely'.

Might it not be useful to have an bit of attention drawn to a conversion 
that is permitted - but may not have been done *correctly* - potentially 
'questionable' IOW?

Not that omnisciance is built into *any* compiler, or there would be no 
Siamese twins.

Even if the GNU people fix 
it, all existing versions of the compiler will complain about it, 
crippling the warnings facility and necessitating dependence upon bugged 
behavior of their -Wcast-qual option to get code to compile 
warnings-free. Really bad. Ironically, g++ does not suffer from this 
problem.

Side question, but does anyone have access to the commercial Intel 8.X 
compiler, can it compile the DragonFlyBSD sources, and what, just for 
comparison, does it do about issues of this sort?

- Not suggesting political change - just curious as to how well gcc 
compares to 'commercial' practice these days.

Bill Hacker



-sigh-






More information about the Submit mailing list