README.md

Be Modular

Prev | Tables of Content | Next

Combining multiple servlets

See diff since last chapter

---

In the last chapter, we read the first line from the TCP connection as the name of the client and send greeting message to the client with its name. Now let's think about how can we make it better.

In fact, parsing the first line seems reusable module and we may want to use it differently. So we can split the original servlet into two, one reads the TCP connection and extract the first line, another reads name and response the greeting message.

The getline servlet

Now we want to make the logic: Getting the first line from input as an independent servlet.So it should have two ports in which reads the input, line which is the first line of the input.

static pipe_t in, out;
static int init(uint32_t argc, char const* const* argv, void* mem)
{
	in = pipe_define("in", PIPE_INPUT, NULL);
	out = pipe_define("line", PIPE_OUTPUT, NULL);
	return 0;
}

And the exec callback is just like the code in the last chapter.

static int exec(void* mem)
{
	char name[128] = {};
	
	size_t count = 0;
	while(!pipe_eof(in))
	{
		size_t read = pipe_read(in, name + count, 1);
		if(name[count] == '\r' || name[count] == '\n') 
			break;
		count += read;
	}

	pipe_write(out, name, count);

	return 0;
}

The say-hello servlet

Instead of just reading the first line, the name port only has the name.

static int exec(void* mem)
{
	char name[128] = {};

	size_t count = pipe_read(in, name, sizeof(name));

	pipe_write(out, what, strlen(what));
	pipe_write(out, ", ", 2);
	pipe_write(out, name, count);

	return 0;
}

Change to the PSS Script

This time we have two servlet, thus we need to change our pipeline definition. We should have two different servlet, one is getline another is say and need to bind the line port of getline servlet with the name port of say.

Service.start({
	read_name := "getline";
	say_hello := @"say Hello";
	() -> "in" read_name "line" -> "name" say_hello "out" -> ();
});