Updates README to describe the API.

Author: Sebmaster

README.md

@@ -4,9 +4,46 @@ thread.js
 Parallel Computing with Javascript
 ---
 
-This branch experiments with a completly rewritten API and internal structure for parallel.js.
+thread.js allows you to spawn webworkers without any hassle in node.js and in the webbrowser.
+The API is really simple:
 
-For details on the implementation look at this [issue](https://github.com/Sebmaster/parallel.js/issues/1).
+var thread = new Thread(data, opts)
+---
+This initializes a new thread.js instance (note: not a webworker). ```data``` is any data you want to work with in your operations.
+In node.js data has to be ```JSON.serialize()```able, in a webbrowser, the object you want to transmit, has to be a transferable.
+
+The following options are available:
+
+* path
+
+  default value: ```(isNode ? __dirname + '/' : '') + 'eval.js'```
+  
+
+  This is the path to the eval.js file on your webserver (webbrowser-only).
+  
+* maxWorkers
+
+  default value: ```isNode ? require('os').cpus().length : 4```
+
+  The maximum of threads you want to use simultaneously.
+
+Note that every function call returns a promise!
+
+.spawn(fn)
+---
+```.spawn``` spawns a single webworker with your given function as an argument.
+The function will receive one parameter with the data which is currently held by the thread.js instance.
+The value returned from spawn will be stored in the thread.js instance after execution in the worker.
+
+.map(fn)
+---
+```.map``` spawns one worker per element in the ```data```-array up to the given ```maxWorkers``` option.
+The function given to map will receive a single element of the array and the returned value will be stored at that index of the array again.
+
+.then(success, fail)
+---
+The functions given to ```.then``` are called after the last requested operation has finished.
+```success``` receives the resulting data object, while ```fail``` will receive an error object.
 
 ### Code Style