This is not a bug in process.mixin, but I think it is undesirable
behavior. Right now process.mixin will not copy over keys with undefined
values. To me that is an unexpected filtering that should not happen
unless specifically called for.

Bug #1 occurred when trying to use process.mixin on a function and
produced a fatal exception.

Bug #2 occurred when trying to do a deep merge with an object containing
one or more objects with a nodeType property. In those cases the deep
copy for this part of the object was not performed and a shallow one was
performed instead.

Both of these bugs were artifacts of the jQuery.extend port.

Change the http.Client API so that it provides a single request() method
taking an optional parameter to specify the HTTP method (defaulting to
"GET"), instead of the five methods get(), head(), post(), del() and put().

- No longer relies on JSON.stringify, so it can output nulls and functions
- Handles circular references better
- Has tests

At the same time implement synchronous wrappers of the POSIX functions.
These will be undocumented until we settle on an API. Works like this

    // returns promise as before
    posix.mkdir("test").addCallback(function () {
      sys.puts("done");
    });

    // returns undefined, executed synchronously.
    posix.mkdirSync("test");
    sys.puts("done");

This refactoring is a step towards allowing promises to be implemented
purely in javascript.

Some people were reporting waf erroring with

  thread.error: can't start new thread

this seems to fix that problem.

http://groups.google.com/group/nodejs/browse_thread/thread/b2edb76691b1848c

Refactored test suite to use the assert module for testing rather than
mjsunit.

added test case for HTTP duplicated header bug on keepalive

http://groups.google.com/group/nodejs/msg/f8e51a8e0c74bd85

Move the event loop calls into javascript.
Makes life so much easier.

Multipart parts kept the first chunk of data after their headers in
memory, even so that was no longer needed.

Reported by onne@onnlucky.com.