I've got a distribution who's job it is to wrap an external C/C++ library. There are dozens of functions I'm wrapping, most are loaded directly, a handful of others have C wrappers in my XS that call the wrapped functions in order to make them work properly with Perl.

In my Test::Pod::Coverage RELEASE_TESTING tests, all of the XS functions are 'bare' (ie. no POD coverage). I don't want to write POD for them, as in each of the Perl subs that present them, I put something like this:

=head2 phys_to_gpio($pin_num)
 
Maps to C<int physPinToGpio(int pin)>
 
Converts the pin number on the physical board to the C<GPIO> represent
+ation,
and returns it.
 
Parameters:
 
    $pin_num
 
Mandatory: The pin number on the physical Raspberry Pi board.
 
=head2 phys_to_wpi($pin_num)
 
Maps to C<int physPinToWpi(int pin)>
 
Converts the pin number on the physical board to the C<wiringPi> numbe
+ring
representation, and returns it.
 
Parameters:
 
    $pin_num
 
Mandatory: The pin number on the physical Raspberry Pi board.
[download]

That allows the user to see what the C call actually is.

So, my question is, how can I avoid Test::Pod::Coverage failures for subs I do not want to cover, because they are XSUBs, and really not used directly?:

t/pod-coverage.t ..... 1/1 
#   Failed test 'Pod coverage on WiringPi::API'
#   at /home/pi/perl5/perlbrew/perls/perl-5.24.1/lib/site_perl/5.24.1/
+Test/Pod/Coverage.pm line 133.
# Coverage for WiringPi::API is 49.6%, with 60 naked subroutines:
#     ads1115Setup
#     analogRead
#     analogWrite
#     bmp180Pressure
...
[download]

All of my Perl subs are snake_case, so anything with camelCase is C/XS.