From: Sam S. <sd...@gn...> - 2011-04-07 14:28:33
|
> * Don Cohen <qba...@vf...3-vap.pbz> [2011-04-06 19:10:34 -0700]: > > This was Bruno's argument :-) great minds work alike > (1) it seems not uncommon to get resets in places where you'd not > expect to complain of a protocol error, such as the case where a > connection is closed without reading all of the input when a wife leaves the room without listening to husband's arguments, I would call that a communication breakdown. > (2) the situation where we currently get econnreset from a read > certainly does meet the definition of eof: > error conditions related to read operations that are done on streams > that have no more data. I think the idea is that EOF is when there are no more data during normal operation; not if there are no more data because the counter-party is dead. > Perhaps the most important consideration is what the cost will be to > programmers with the two alternatives. you are assuming that the users will want to treat ECONNRESET the same way as EOF (end of communication). I think the right way to treat econnreset is to restart the communication and find out what happened to the counter-party. -- Sam Steingold (http://sds.podval.org/) on CentOS release 5.5 (Final) X http://ffii.org http://thereligionofpeace.com http://camera.org http://pmw.org.il http://honestreporting.com (let ((a "(let ((a %c%s%c)) (format a 34 a 34))")) (format a 34 a 34)) |