Re: Additional GCC warnings?

[Ross Smith <r-smith at ihug dot co dot nz>]

On Friday, 21 March 2003 07:31, David Taylor wrote:
> EMC has, in house, added several new warnings to GCC.  Are people
> generally in favor of adding new warnings to GCC?  If yes, then
> assuming the implementation/etc is acceptable, which of the following
> would people be interested in seeing patches for?
>
> -Waggregate-passed-by-value / -Wno-aggregate-passed-by-value
>
>     Warn if an aggregate parameter is passed by value rather than
>     by reference.

Not interested.

> -Wenum-mismatch / -Wno-enum-mismatch
>
>     Warn when value assigned to a variable of an enumeral type
> doesn't correspond to an enumerator

This would be useful.

> -Wprecision-mismatch / -Wno-precision-mismatch
>
>     Warn when the precision of the destination is smaller than
>     the precision of the source.

This would be _very_ useful! Do you intend it to include integers, or 
only floating point? FP alone would be useful, but it would be even 
more useful with integers.

What about signed/unsigned and int/float conversions? GCC currently 
warns about some mixed-mode arithmetic, but not all.

> -Winclude-depth-exceeds=<n>
>
>     Warn if the number of nested levels of #include'd files exceeds
>     <n>.  Warning can be suppressed by #pragma
> include_depth_check_off and re-enabled by #pragma
> include_depth_check_on.

Not interested. (Actually, I can't imagine what this could be useful to 
anyone for.)

> -Wpointer-cast-to-signed / -Wno-pointer-cast-to-signed
>
>     Warn about any casting of pointer type data to a signed entity

Possibly useful.

-- 
Ross Smith ......... r-smith at ihug dot co dot nz ......... Auckland, New Zealand

Eagles may soar, but weasels never get sucked into a land war in Asia.