Expressive middleware for node.js using generators via [co](https://github.com/visionmedia/co)
to make writing web applications and REST APIs more enjoyable to write.
Only methods that are common to nearly all HTTP servers are integrated directly into Koa's small ~400 SLOC codebase. This
includes things like content-negotiation, normalization of node inconsistencies, redirection, and a few others.
No middleware are bundled with koa. If you prefer to only define a single dependency for common middleware, much like Connect, you may use
[koa-common](https://github.com/koajs/common).
## Installation
```
$ npm install koa
```
To use Koa you must be running __node 0.11.4__ or higher for generator support, and must run node(1)
with the `--harmony-generators` flag. If you don't like typing this, add an alias to your shell profile:
```
alias node='node --harmony-generators'
```
## Community
- [API](docs/api.md) documentation
- [Middleware](https://github.com/koajs/koa/wiki/Middleware) list
- [Wiki](https://github.com/koajs/koa/wiki)
- [G+ Community](https://plus.google.com/communities/101845768320796750641)
- [Mailing list](https://groups.google.com/forum/#!forum/koajs)
- __#koajs__ on freenode
## Example
```js
var koa = require('koa');
var app = koa();
// logger
app.use(function(next){
return function *(){
var start = new Date;
yield next;
var ms = new Date - start;
console.log('%s %s - %s', this.method, this.url, ms);
}
});
// response
app.use(function(next){
return function *(){
yield next;
this.body = 'Hello World';
}
});
app.listen(3000);
```
## Running tests
```
$ make test
```
## Benchmarks
Average latency with one middleware:
0.628ms
Average latency with __400__ noop middleware:
1.5ms
If you like silly benchmarks, here's the requests per second:
```
1 middleware
8083.45
5 middleware
7449.95
10 middleware
7166.66
15 middleware
6992.18
20 middleware
6650.28
30 middleware
6113.57
50 middleware
5117.46
```
With __50__ middleware (likely much more than you'll need), that's __307,020__ requests per minute, and __18,421,200__ per hour, so unless you're a facebook and can't manage to spin up more
than one process to scale horizontally you'll be fine ;)
# License
MIT