Hi, "man 2 accept" says that the socket created by accept() has the same properties as the socket which was used in the accept call. Is it possible that in Java the created socket should have default properties here instead? That means we should not only reset the timeout but other socket options as well. Ideas? cya Robert Mark Wielaard wrote: > Hi Edwin, > > On Wed, 2006-05-03 at 16:34 +0200, Edwin Steiner wrote: > >>I tracked down a problem where JOnAS on classpath/cacao dropped >>a database connection for no good reason. The cause was that the >>ServerSocket used for "accept" has a one-second timeout set, and >>the accepted Sockets inherit it. >> >>This inheritance of options is documented for the accept(2) >>system call, but I don't think it is intended behaviour on >>the java.net.Socket level. I did not find any reference to >>that in the ServerSocket documentation. > > > Good catch. I fixed it in CVS (and on the 0.91 release branch). > And I added a Mauve test for this. > > Thanks, > > Mark -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 252 bytes Desc: OpenPGP digital signature Url : http://developer.classpath.org/pipermail/classpath/attachments/20060515/b09ce971/signature.pgp
