Danny Niu wrote, on 22 Jul 2022: > > How did we decide to resolve bug 1068 by accepting it as marked? > Did we do some liaison work with IETF? Surveyed existing > implementations? Or both, or anything else?
Effectively we surveyed existing implementations, although it wasn't a formal survey. Nobody knows of any implementation of the sockets API that doesn't do it. It is even documented behaviour in MS Windows and IBM z/OS. The bug should have been resolved, in the way it now has been, when it was first discussed in a teleconference some years ago, but somehow we let Mark Z (aka "shwaresyst") persuade us to defer it. His arguments against it have been clearly bogus all along and it is ridiculous that it took this long for us to dismiss them. The timing of the resolution is just because we are in the process of going through old open bugs and it was the next in the sequence. > > 2022-07-22 21:29:10，Andrew Josey via austin-group-l at The Open Group > > <austin-group-l@opengroup.org> 写道： > > > > All > > Enclosed are the minutes from the Thursday meeting this week > > regards > > Andrew > > -------------- > > > > Minutes of the 21st July 2022 Teleconference Austin-1241 Page 1 of 1 > > Submitted by Andrew Josey, The Open Group. 22nd July 2022 > > > > Attendees: > > Don Cragun, IEEE PASC OR > > Nick Stoughton, Logitech/USENIX, ISO/IEC JTC 1/SC 22 OR > > Andrew Josey, The Open Group > > Geoff Clare, The Open Group > > Eric Blake, Red Hat, The Open Group OR (partial) > > Mark Ziegast, SHware Systems Dev. > > Tom Thompson, IEEE > > Eric Ackermann, HPI, University of Potsdam > > > > > > * General news > > > > This was a call dedicated to general bugs. > > > > * Current Business > > > > Bug 1068: Binding to a system-assigned port Accepted as Marked > > <mangled url redacted> > > > > This item is tagged for Issue 8 > > > > On D2.1 page 298 line 10373 section <netinet/in.h>, after: > > The sin_port and sin_addr members shall be in network byte order. > > add: > > If the sin_port value passed to bind() is zero, the port number > > bound to the socket shall be one chosen by the implementation > > from an implementation-defined port range to produce an unused > > local address. > > > > On D2.1 page 298 line 10393 section <netinet/in.h>, after: > > The sin6_port and sin6_addr members shall be in network byte order. > > add: > > If the sin6_port value passed to bind() is zero, the port number > > bound to the socket shall be one chosen by the implementation > > from an implementation-defined port range to produce an unused > > local address. > > > > On D2.1 page 301 line 10510 section <netinet/in.h>, add bind() to SEE ALSO -- Geoff Clare <g.cl...@opengroup.org> The Open Group, Apex Plaza, Forbury Road, Reading, RG1 1AX, England
