[dmd-concurrency] Shutdown protocol

[Michel Fortin]

Le 2010-01-20 à 14:31, Andrei Alexandrescu a écrit :

> Good example. Is the interim conclusion to leave things as they are at least for the time being? I don't think they make things any more complicated than models that bring their own complications, and the current API does make a great deal of cases much simpler.

For the time being we could leave things as they are, but add the capability to manually call shutdown in a simple way. The example at the start of this thread would thus be rewritten like this, where I added "wait" in main and "scope (exit) tid.shutdown()" in copy:

	import std.algorithm, std.concurrency, std.stdio;

	void main() {
		wait(
			spawn(&copy("in1", "out1")),
			spawn(&copy("in2", "out2"))
		);
	}

	void copy(string input, string output) {
		File source = File(input, "r");

		enum bufferSize = 1024 * 100;
		auto tid = spawn(&writer, outputPath); // Read loop
		scope(exit) tid.shutdown();
		foreach (immutable(ubyte)[] buffer; stdin.byChunk(bufferSize)) {
			send(tid, buffer);
		}
	}

	void writer(string output) {
		File target = File(output, "w");

		// Write loop
		for (;;) {
			auto buffer = receiveOnly!(immutable(ubyte)[])();
			target.rawWrite(buffer);
		}
	}

The only thing worrying me a little is that it's easy to forget calling tid.shutdown() when an exception is thrown in the copy thread. It's really important to use of scope(exit) or some similar mechanism -- except in short demo programs like the one above unfortunately -- to avoid zombie threads.

I can't help but notice that the code above is quite like manual memory management, except with threads. Perhaps what we need is a GC for unreferenced threads...


-- 
Michel Fortin
michel.fortin at michelf.com
http://michelf.com/