Hi all,
I'm in the process of learning pharo. As an exercise I'm implementing a "Inter-VM Object channel": - objects serialized/deserialized via Fuel - bidi channel connects 2 vm's - sendObject and receiveObject as api - transport configurable as either Socket or SharedMem (as an opportunity to learn the threaded FFI stuff) During the implementation of the server socket side, I stumbled over a strange behavior when the backlog is set to a value less than 2: Here is a simplified version of the server side: listeningSocket := Socket newTCP. listeningSocket listenOn: 19999 backlogSize: 1. "NOTE the 1 here!" clientSocket := listeningSocket waitForAcceptFor: 30. ... When a client connects: echo Hallo | nc -v 127.0.0.1 19999 listeningSocket waitForAcceptFor: is unblocked and signals an error which I think is then masked by another error which is: "SocketError: cannot access socket error code" which is raised from Socket primSocketError: Note that no error is signaled as soon as the backlog size is 2 or more. - Is there a reason for this behavior or is this a bug? - Is my analysis correct that the original error is hidden by another error that is raised during handling of the original error? Regards, Martin -- ____________________________________ Martin Fischer
