Stupidly easy flow control.
While kgo still works fine, I've since moved my efforts over to the much cleaner, faster, better tested, more flexible, and better maintained, righto, to which you can easily and incrementally migrate to if you want.
flow contol should be seamless, you should be able to say what you want done, and say kgo.
kgo(result name, [dependencies], asynchronous function); // -> self
where result name
is an arbitrary string that represents the result of the function,
dependencies
is an array of strings that map to the result of another function,
and asynchronous function
is a function that, when complete, calls a callback with it's results.
kgo returns its-self, so it can be chained:
kgo
(name, deps, fn)
(name, deps, fn)
(name, deps, fn);
require kgo:
var kgo = require('./kgo');
use kgo:
kgo
('things', function(done){
//Something async
setTimeout(function(){
done(null, 1);
}, 100);
})
('stuff', function(done){
//Something async
setTimeout(function(){
done(null, 2);
}, 100);
})
('whatsits', ['things', 'stuff'], function(things, stuff, done){
//Something async
setTimeout(function(){
done(null, things + stuff);
}, 100);
})
('dooby', ['things'], function(things, done){
//Something async
setTimeout(function(){
done(null, things/2);
}, 100);
})
(['whatsits', 'dooby'], function(whatsits, dooby, done){
//Done
console.log(whatsits, dooby);
});
The above will log 3, 0.5;
Removed as of version 2. Use foreign instead.
You will often not need the result of a dependency, and it's annoying to have unused parameters in your functions. You can specify that you have a dependency, whos result you don't want, by prefixing the dependency name with an exclamation mark:
kgo
('a', function(done){
done(null, 'foo');
})
('b', ['!a'], function(done){
done(null, 'bar');
})
(['b'], function(b){
// here b will be "bar"
});
done
function:kgo
('foo', getFoo) // getFoo will be passed (callback)
('bar', ['foo'], getFoo) // getFoo will be passed (bar, callback)
(['*', 'bar'], function(error, bar){
// arguments.length will be 2
})
You can define default data for use in later tasks by passing an object into kgo, where the keys in the objects will map to dependency names:
kgo
({
foo: 1
})
('bar', function(done){
done(null, 2);
})
('baz', ['foo', 'bar'], function(foo, bar, done){
});
This is especially useful when you want to use named functions that need additional parameters to run:
var fs = require('fs');
kgo
({
'sourcePath': '/foo/bar'
})
('files', ['sourcePath'], fs.readdir);
You can return more than one result in a single task by giving your task multiple names, and returning more results in the callback
kgo
('foo', 'bar', function(done){
done(null, 2, 4);
})
('baz', ['foo', 'bar'], function(foo, bar, done){
// foo === 2
// bar === 4
});
Sometimes you want to just run a synchonous function as a step in kgo, to do this you can use kgo.sync
, which will return a callback-passing-style version.
kgo
('task1', task1);
('task2', ['task1'], kgo.sync(function(x){ return x + 1 }))
('task3', ['task2'], task3);
You can handler errors from specific tasknames by prefixing the taskname with an astrix (*)
any task that has a dependency on any error task will resolve if any error dependency is resolved.
kgo
('task1', task1)
('task2', task2)
(['*task1'], function(tast1error){
// Called if task1 errored
})
(['*task2'], function(tast2error){
// Called if task2 errored
})
(['task1', 'task2'], function(result1, result2){
// Called if task1 and task2 succeded.
});
there is an implicit error task, *
, that will resolve to the first of any error that occurs.
kgo
(task)
(another task)
(['*'], function(error){
// Called if any error occurs in any task
});
You can combine this with other dependencies to build a callback task:
function doThings(callback){
kgo
('task1', ...)
('task2', ...)
('result', ['task1', 'task2'], ...)
(['*', 'result'], callback); // Will be called either if any task errors, OR if result resolves.
}