[bug-gawk] getaddrinfo() in socketopen() (io.c) fails for AF_INET[46] -- unnecessarily - CORRECTION

[Houder]

On 2018-02-08 18:59, Houder wrote:
[snip]

> static int
> socketopen(int family, int type, const char *localpname,
>         const char *remotepname, const char *remotehostname, bool 
> *hard_error)
> {
> [snip]
>
>         memset(& lhints, '\0', sizeof (lhints));
>
> > >         /* if only the loopback interface is up and hints.ai_flags 
> > > has AI_ADDRCONFIG,
> > >            getaddrinfo() will return the addresses of the loopback 
> > > interface, but only
> > >            if hints.ai_family == AF_UNSPEC
> > >         */
>         lhints.ai_flags = AI_PASSIVE | AI_ADDRCONFIG;
>
>         lhints.ai_socktype = type;
>         lhints.ai_family = family;
>
> > > // do return the loopback address in case the loopback interface is 
> > > the only one that
> > > // is up (and hints.ai_family == AF_INET[46])
> > > if ( (lhints.ai_family == AF_INET) || (lhints.ai_family == AF_INET6) 
> > > )
> > >    lhints.ai_flags = AI_PASSIVE;
>
>         lerror = getaddrinfo(NULL, localpname, & lhints, & lres);
> ...

Sorry, I should have corrected my comments in gawk-4.2.0/io.c before 
sending
my e-mail ...

1.
/* if only the loopback interface is up and hints.ai_flags has 
AI_ADDRCONFIG,
   getaddrinfo() will succeed and return all wildcard addresses, but 
only if
   hints.ai_family == AF_UNSPEC
 */

2.
// do return the wildcard address in case the loopback interface is the 
only
// one that is up (and hints.ai_family == either AF_INET4 or AF_INET6)

Henri