comparison src/protocols/oscar/ft.c @ 9430:30a812571a6b

[gaim-migrate @ 10250] I haven't tested this, but it should be ok. Read my note. Also, I'm using PF_INET instead of AF_INET for the reasons given at http://www.ecst.csuchico.edu/~beej/guide/net/html/syscalls.html#socket (This is one of those things were it doesn't matter but it feels more correct.) committer: Tailor Script <[e-mail]>

9429:a3e3c71d995a

#ifdef _WIN32
#include "win32dep.h"
#endif

struct aim_odc_intdata {
	fu8_t cookie[8];
	char sn[MAXSNLEN+1];
	char ip[22];
};

	unsigned short port;

	if ((acceptfd = accept(cur->fd, &addr, &addrlen)) == -1)
		return 0; /* not an error */

	/* Also accept inet6? */
	if (addr.sa_family != AF_INET) {
		close(acceptfd);
		aim_conn_close(cur);
		return -1;
	}

9430:30a812571a6b

#ifdef _WIN32
#include "win32dep.h"
#endif

/*
 * I really want to switch all our networking code to using IPv6 only,
 * but that really isn't a good idea at all.  Evan S. of Adium says
 * OS X sets all connections as "AF_INET6/PF_INET6," even if there is
 * nothing inherently IPv6 about them.  And I feel like Linux kernel
 * 2.6.5 is doing the same thing.  So we REALLY should accept
 * connections if they're showing up as IPv6.  Old OSes (Solaris?)
 * that might not have full IPv6 support yet will fail if we try
 * to use PF_INET6 but it isn't defined.  --Mark Doliner
 */
#ifndef PF_INET6
#define PF_INET6 PF_INET
#endif

struct aim_odc_intdata {
	fu8_t cookie[8];
	char sn[MAXSNLEN+1];
	char ip[22];
};

	unsigned short port;

	if ((acceptfd = accept(cur->fd, &addr, &addrlen)) == -1)
		return 0; /* not an error */

	if ((addr.sa_family != PF_INET) && (addr.sa_family != PF_INET6)) {
		close(acceptfd);
		aim_conn_close(cur);
		return -1;
	}