Re: cvs: php4 /ext/sockets sockets.c

I am equally surprised that this fix works, though I don't doubt your discovery. I have a few questions though What version/sp did you see this leak on? What SAPI? Was recv the only call that leaked? Thanks, Jason On Sat, 2002-06-15 at 10:58, Harald Radi wrote: > phanto Sat Jun 15 11:58:20 2002 EDT > > Modified files: > /php4/ext/sockets sockets.c > Log: > workaround for a win32 memleak > # this definitely has to be revisited ! > # maybe we should use winsock2 and overlapped IO > > Index: php4/ext/sockets/sockets.c > diff -u php4/ext/sockets/sockets.c:1.114 php4/ext/sockets/sockets.c:1.115 > --- php4/ext/sockets/sockets.c:1.114 Sun May 12 03:22:47 2002 > +++ php4/ext/sockets/sockets.c Sat Jun 15 11:58:16 2002 > @@ -19,7 +19,7 @@ > +----------------------------------------------------------------------+ > */ > > -/* $Id: sockets.c,v 1.114 2002/05/12 07:22:47 jason Exp $ */ > +/* $Id: sockets.c,v 1.115 2002/06/15 15:58:16 phanto Exp $ */ > > #ifdef HAVE_CONFIG_H > #include "config.h" > @@ -56,7 +56,6 @@ > # define set_errno(a) (errno = a) > # define set_h_errno(a) (h_errno = a) > #else /* windows */ > -# include <winsock.h> > # include "php_sockets.h" > # include "php_sockets_win.h" > # define IS_INVALID_SOCKET(a) (a->bsd_socket == INVALID_SOCKET) > @@ -755,6 +754,8 @@ > retval = (*read_function)(php_sock->bsd_socket, tmpbuf, length); > #else > retval = recv(php_sock->bsd_socket, tmpbuf, length, 0); > + /* i don't know why, but it _does_ fix a memleak */ > + SleepEx(1, TRUE); > #endif > > if (retval == -1) { > > > > -- > PHP CVS Mailing List (http://www.php.net/) > To unsubscribe, visit: http://www.php.net/unsub.php >