Understanding compiler warning options

[Edward Diener]

On 10/1/2015 6:22 PM, Manuel López-Ibáñez wrote:
> On 30/09/15 03:57, Martin Sebor wrote:
>> On 09/29/2015 05:41 PM, Edward Diener wrote:
>>> I am trying to understand how the compiler decides which warnings are
>>> issued when compiling a translation unit.
>>>
>>> What is the default warnings that are issued if no compiler warning
>>> options is set on the command line ?
> [...]
>> An example might be the -Wdiv-by-zero warning that's defined like
>> this:
>>
>>    Wdiv-by-zero
>>    C ObjC C++ ObjC++ Var(warn_div_by_zero) Init(1) Warning
>>    Warn about compile-time integer division by zero
>
> Unfortunately, many options are enabled under some conditions and this
> cannot (yet) be encoded in the .opt files: https://gcc.gnu.org/PR53063
>
> Any contribution in that direction would be welcome.
>
>> You may be able to tell (probably less reliably) from the manual
>> by searching for warning options listed in the negative -Wno-xxx
>> form. For instance:
>>
>>    -Wno-endif-labels
>>        Do not warn whenever an #else or an #endif are followed by text.
>>
>> means the option is enabled by default. Other options may say
>> when (e.g., in what language) they're enabled by default.
>
> I think the consensus nowadays is that options that are enabled by
> default under some condition should be documented as such directly.
> Documenting the -Wno- form is not really useful as a substitute for this
> because the "default" depends on other options (e.g., -std=), language,
> target, etc. Documentation patches welcome!
>
>>> If I set a general option, such as -w, -Wall, or -Wextra, do these get
>>> overridden by more specific compiler options no matter where in the
>>> command line they are encountered, or does the order of the compiler
>>> warnings on the command line matter in such a case ? As a specific
>>> practical example if the command line has "-Wno-unused-local-typedef
>>> -Wall ..." does the "-Wno-unused-local-typedef" override the "-Wall"
>>> option ?
>>
>> Unfortunately, this is neither documented nor necessarily intuitive
>> (YMMV). IME, the best way to find out is by experimenting.
>
> It is documented here:
> https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options
>
> but perhaps it could be better documented. Again, unless someone gives
> it a try to improve the documentation, it will remain the same.
>
> In particular: "You can request many specific warnings with options
> beginning with ‘-W’, for example -Wimplicit to request warnings on
> implicit declarations. Each of these specific warning options also has a
> negative form beginning ‘-Wno-’ to turn off warnings; for example,
> -Wno-implicit. This manual lists only one of the two forms, whichever is
> not the default.
>
> Some options, such as -Wall and -Wextra, turn on other options, such as
> -Wunused, which may turn on further options, such as -Wunused-value. The
> combined effect of positive and negative forms is that more specific
> options have priority over less specific ones, independently of their
> position in the command-line. For options of the same specificity, the
> last one takes effect."

The next to last line you quoted implies to me that the "-w" option can 
be overriden by nore specific warning messages which turn back on a 
particular warnings, since "-w" is a general way to turn off all warnings.

>
> The above behavior is handled automatically nowadays such that all
> options behave the same. If it doesn't work, something is wrong in the
> .opt file or the option was never converted to the automatic way of
> handling options dependencies and it is being set explicitly somewhere,
> which is also a latent bug.