Is there any reasonable initial/invalid value for a pid_t?

Turning on the -Wextra flag to gcc seems to have the effect of disallowing partial initialization of structs. E.g.:

// main.c
#include <pthread.h>

typedef struct S {
  int i;
  pid_t pid;
} S;

int main( int argc, char* argv[] ) {
  (void)argc;
  (void)argv;
  S s = { 42 };
  (void)s;

  return 0;
}

$ gcc --version && gcc -Wall -Wextra -pedantic -Werror ./main.c
gcc (Debian 6.3.0-18+deb9u1) 6.3.0 20170516
Copyright (C) 2016 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

./main.c: In function ‘main’:
./main.c:12:3: error: missing initializer for field ‘pid’ of ‘S {aka struct S}’ [-Werror=missing-field-initializers]
   S s = { 42 };
   ^
./main.c:6:9: note: ‘pid’ declared here
   pid_t pid;
         ^~~
cc1: all warnings being treated as errors

Is there any reasonable initial (invalid) value for a pid_t? I'd always felt that since pid_t is opaque, one ought not make any assumptions about flag values.

If there is no reasonable initial/invalid value for a pid_t is there some other good/widely-used programming practice to not hit this error?

Note: specifically interested in learning if there are options that do not involve assigning an arbitrary value that is assumed invalid until a separate validity flag variable is set true, e.g.

typedef struct S {
  int i;
  pid_t pid;
  bool valid;
} S;

S s = { 42, 9, false };

pid_t is not opaque. Zero and all negative values are explicitly exceptional (used e.g. by waitpid to represent particular classes of processes to wait for) and arguably 1 is exceptional too since the specialness of -1 prevents 1 from being a process group id you can use normally (traditionally, pid 1 is the init process).

For your purpose, 0 seems like the most reasonable choice.