add msg id NULL check for normal ops

NEWS.md

@@ -1,11 +1,12 @@
-# 0.8.10
+# 0.8.95
 
 * We are now using _ZeroMQ_ via `Rcpp` in preparation for `v0.9` (#151)
 * New `multiprocess` backend via `callr` instead of forking (#142, #197) FIXME: test case
 * Sending data on sockets is now blocking to avoid excess memory usage (#161)
 * `multicore`, `multiprocess` schedulers now support logging (#169)
 * New option `clustermq.host` can specify host IP or network interface name (#170)
 * Template filling will now raise error for missing keys (#174, #198)
+* Workers failing with large common data is improved (fixed?) (#146, #179, #191)
 * Local connections are now routed via `127.0.0.1` instead of `localhost` (#192)
 * Submit messages are different between local, multicore and HPC (#196)
 * Functions exported by `foreach` now have their environment stripped (#200)

R/worker.r

@@ -39,6 +39,11 @@ worker = function(master, timeout=getOption("clustermq.worker.timeout", 600),
         if (events[1]) {
             tic = proc.time()
             msg = zmq$receive()
+            if (is.null(msg$id)) {
+                # more information if #146, #179, #191 happen again
+                message("msg: ", msg)
+                next
+            }
             delta = proc.time() - tic
             message(sprintf("> %s (%.3fs wait)", msg$id, delta[3]))
         } else