+1 to using an existing socket, it's more flexible than listen(0) + report as per the inetd case. qpidd started out doing listen(0) + report and we ended up having to do the existing socket thing as well - just existing socket covers both with less code in the proactor.
On Fri, Oct 20, 2017 at 3:29 PM, Robbie Gemmell (JIRA) <[email protected]> wrote: > > [ https://issues.apache.org/jira/browse/PROTON-1646?page= > com.atlassian.jira.plugin.system.issuetabpanels:comment- > tabpanel&focusedCommentId=16212712#comment-16212712 ] > > Robbie Gemmell commented on PROTON-1646: > ---------------------------------------- > > Another common approach for tests can be for a component to be told to > bind a free port by itself, with it then offering a way to find out what it > was. > > > Proactor API needs a way to use an existing socket. > > --------------------------------------------------- > > > > Key: PROTON-1646 > > URL: https://issues.apache.org/jira/browse/PROTON-1646 > > Project: Qpid Proton > > Issue Type: Improvement > > Components: cpp-binding, proton-c > > Affects Versions: proton-c-0.18.0 > > Reporter: Andrew Stitcher > > Assignee: Cliff Jansen > > > > There are various times when it is necessary to take an existing socket > and use it to communicate. > > * The most obvious is when we are started by systemd or inetd and passed > the listening socket that they have accepted on our behalf. > > * Another case is in tests when we might want to open a port using a > random port number and then atomically hand that to test code. > > > > -- > This message was sent by Atlassian JIRA > (v6.4.14#64029) > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
