Node Roundup 4

2010-08-25 00:00:00 +0100 by Alex R. Young

Welcome to the Node Roundup. Send your apps and libraries in for review
through our contact form or


Webshell from Fictive Kin (by Evan Haas and Sean Coates) is a
JavaScript HTTP client, released under Apache License 2.0 that uses

You can issue commands like this:

It supports HTTP auth, cookies, and JSON. There's lots of examples of
usage in the README. It seems like something I might end up using when
I'm debugging JSON-heavy Node projects. And by "debugging" I mean
"should be writing tests but would prefer to do things the hackish way."


nDistro by TJ Holowaychuk is a way of quickly installing Node with a bunch of dependencies, based on
simple configuration files. Configuration looks like this:

node 0.1.102
module senchalabs connect
module visionmedia express 1.0.0beta2
module visionmedia connect-form
module visionmedia connect-redis
module visionmedia jade
module visionmedia ejs

This is useful for a few things: testing your libraries against
different versions of Node, or recreating a particular environment for a
production server. It's a shell script that downloads binary
distributions of Node, which means you should technically only need a
shell to get Node up and running.

Full installation and usage instructions are available in the

The author also posted a screencast of nDistro
in use.

Node on Windows

If you usually work on a Windows desktop machine, you might have found
building Node a bit of a chore. Try following Step by step instructions
to install NodeJS on Windows
-- it
relies on Cygwin but at least you can get Node running without
dual-booting or starting a VM.

Japanese Node Group

I noticed there's a Japanese Node Google
started by Toshihiro
Shimizu. He's also been translating the Node site into Japanese at

Group Discussions

V8 Object Limits,
started by Timothy Caswell, is an interesting discussion on the nodejs
Google Group about performance and plain JavaScript objects. He was
looking into the speed at which V8 can insert properties, then analysing
the performance characteristics as the object grows.

You'll notice that it starts out nice and fast at 2,673,179 inserts per second, but by the time it reaches about 13,273,357 properties, it's down to taking almost 2 seconds to insert a single property. Also I'm seeing some blips in my graph that I assume are the stop-the-world garbage collector.

A related discussion is Node / v8 1gb memory

TJ Holowaychuk mentioned that perhaps frameworks should adopt
NODE_ENV as a convention for determining the application's environment
(production, development, testing, etc.) I think this would be useful, especially now that everyone on GitHub has started their own Node web