Thanks for those references. That makes sense. On Sun, Jul 21, 2024 at 8:04 PM Gregory Nutt <spudan...@gmail.com> wrote:
> > > > > On 7/21/2024 5:47 PM, Gregory Nutt wrote: > >> > >> On 7/21/2024 5:38 PM, Nathan Hartman wrote: > >>> Hi Alan, > >>> > >>> I think we should have "int main(int argc, const char * argv[])" (added > >>> 'const' for argv). Maybe should grep for all instances of main in > >>> our code > >>> and docs and add const? I'm willing to do that. > >>> > >>> Let me know, > >>> Nathan > >> > >> POSIX specifies the required prototype of main at > >> https://pubs.opengroup.org/onlinepubs/007904975/functions/exec.html > >> Yes, main() is specified in exec() and other locations refer to the > >> exec() page for the specification of main(). > >> > >> POSIX does specify the prototype from the standpoint of the OS call > >> into the application. Applications may, of course, use any form that > >> they like but the OS must conform to this standard. > > > > Also interesting: > > > https://stackoverflow.com/questions/5808679/why-is-main-argument-argv-of-type-char-rather-than-const-char > > That reference explains where argv must not be const. Basically, it > > is not const because it should be modifiable. > > > > Not as good, but to the same point: > > https://stackoverflow.com/questions/20558418/why-is-argc-not-a-constant > With at least some compilers you will be an error if argv is const > becuase of pointer mismatch. This is expected POSIX-compliant behavior: > > https://cboard.cprogramming.com/c-programming/173878-making-const-char-*-out-argv.html > > > > > >> > >> >
