git.iko.soy/distributed-process
on 78147be help sign in
78147beCleanupEdsko de Vries b7aacf8Test that CloseSocket requests can be ignoredEdsko de Vries 8690391Reenable the early disconnect testEdsko de Vries 528cfd8Add a stres test for parallel calls to connectEdsko de Vries d063618Fix the CloseReopen testEdsko de Vries 2e91bf6Remove some racy conditions (one still left)Edsko de Vries 3e91243Close TCP sockets when no longer requiredEdsko de Vries 83cc27bExpose TCP transportEdsko de Vries 70fa7e7Add refcount to endpointEdsko de Vries 7a0b324Fix the test (in-memory transport was fine)Edsko de Vries 7b1d69bExpose a TODO in the TCP transportEdsko de Vries 478d769Continue work on error handlingEdsko de Vries c067c47More error handlingEdsko de Vries 1fe6a62Remove obsolete commentEdsko de Vries c011655Lightweight connections/bidirectional socketsEdsko de Vries 55b8cd8Reuse sockets for connectionsEdsko de Vries fdf266fCleanupEdsko de Vries 85860aeUse 32-bit integres for connection IDsEdsko de Vries bf6c7a3Send ConnectionClosed when connection is brokenEdsko de Vries 39d3f1eDeal with errors during sendingEdsko de Vries 3938dfcReorganize code a bitEdsko de Vries dfcf360TCP layer passes unit testsEdsko de Vries 3217273Basic implementation of the TCP transportEdsko de Vries 53e0889Add multicast support to in-memory transportEdsko de Vries 386ead7Modify multicast API, test the in-memory transportEdsko de Vries 583332dStart work on slightly alternative Transport APIEdsko de Vries a906652Use MVar rather than Chan in Transport.TCPEdsko de Vries dd9b0b4Start documentationEdsko de Vries ec9029fTCP_NODELAY does not seem to improve thingsEdsko de Vries a366d5eCompare one socket pair with twoEdsko de Vries c7f72ceWhoops, left some debugging code inEdsko de Vries 610efb9Start comparing the transport against base testsEdsko de Vries 842cc13Indirection (Chan/MVar) does not cause slowdownEdsko de Vries 5674553SkeletonEdsko de Vries 212352cRename fileEdsko de Vries a050b8aFast (de)serialization of intsEdsko de Vries 8049006Don't set forking client separatelyEdsko de Vries bc4312eTwo more testsEdsko de Vries f6240c3Single receive makes no differenceEdsko de Vries 99fe9c4Add header, see what slows downEdsko de Vries 9991818Turns out that -threaded is the culpritEdsko de Vries 636c94eLarger sample sizeEdsko de Vries 4e2c09fBarebones ping test. Haskell approximates C.Edsko de Vries 94a1dd5Minor modifications to improve profilingEdsko de Vries 90cb954Run server and client inside a single processEdsko de Vries 39d8085Modify the benchmarks to output the latenciesEdsko de Vries cb07733Compute average and stddev of the round-trip latencyEdsko de Vries ee77eebC version of the ping testEdsko de Vries 555b2c1Extend test to show bug in pipes transportEdsko de Vries 6b26fa9fix: I had checked in with some DemoTransport transports disabledRyan Newton