>>>>> On Sun, 31 Jul 2011 09:04:42 +0200, Bart Van Assche <[email protected]> >>>>> said:
BVA> - As far as I know setting MSG_NOSIGNAL only makes sense for stream BVA> (connected) sockets and not for UDP sockets, so it should be fine to remove BVA> all uses of MSG_NOSIGNAL in snmplib/transports/snmpUDPBaseDomain.c. I'll defer to someone that knows the networking socket code on all the platforms :-/ BVA> - In apps/sshtosnmp.c using MSG_NOSIGNAL makes sense but is nonportable. A BVA> possible approach here is to remove MSG_NOSIGNAL and make the OS ignore BVA> SIGPIPE. That should help on OS/X where MSG_NOSIGNAL is not available but BVA> where SIGPIPE can be generated by the kernel. We could always leave it and trigger on SIGPIPE too. I'm not convinced everywhere will generate a SIGPIPE? Does windows do that properly? I'd assume so since it's you that's posting the suggestion. BVA> - The use of MSG_DONTWAIT in apps/sshtosnmp.c looks dangerous to me BVA> since there is no guarantee that all data gets sent if this flag is BVA> specified. How about removing that flag and setting the socket BVA> option SO_SNDTIMEO instead ? At least on Linux that option works BVA> for Unix sockets. The data will all get sent or the session will break at some point in the future. I'm not sure I agree it's dangerous. I'm not sure SO_SNDTIMEO is portable either (based on some quick (lame) header searches). We could, again, do both? -- Wes Hardaker Please mail all replies to [email protected] ------------------------------------------------------------------------------ BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA The must-attend event for mobile developers. Connect with experts. Get tools for creating Super Apps. See the latest technologies. Sessions, hands-on labs, demos & much more. Register early & save! http://p.sf.net/sfu/rim-blackberry-1 _______________________________________________ Net-snmp-coders mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
