Release Notes: The algorithm for cycle detection (paexec -s) was completely reworked and now does't allocate tasks_count^2 integers. Now it works much faster. The -Z _timeout_ option was added. An attempt to rerun a command on a failed node is made every _timeout_ seconds. This option makes it possible to organize clusters over unreliable networks, hardware or software.
Release Notes: This release adds a -z option. If applied, read/write(2) operations from/to nodes become non-critical. In case paexec has lost connection to the node, it will reassign the failed task to another node and, if the -s option is applied, will output the string "fatal" to stdout. This makes paexec resistant to the I/O errors, and as a result you can create paexec clusters even over a network consisting of unreliable hosts (Internet?). Failed hosts are marked as such, and will not be used during the current run of paexec. There are minor fixes.
Release Notes: This release adds a -s option which causes a partially-ordered set of tasks to be processed. Instead of autonomous tasks, a graph of the tasks is read from stdin and processed. In this mode, every task can either FAIL or SUCCEED. Processing any given task is started if and only if all its dependencies succeeded. Failure in dependency causes recursive failures.