"Bezel"  bezel.rb

The idea of Bezel is to overcome the limitations of using different versions of the same package in the same Ruby proccess.

It works like this. Let’s say I wrote a library called TomsLib. Now I want to use TomsLib in my new fancy app, FancyApp. In my FancyApp namespace I have to create a reference to TomsLib.

module FancyApp
  TomsLib = lib('tomslib', '1.5')
  ...

Now I have access to TomsLib, but it is localized to my application. If Jane comes along and wants to use a different version of TomsLib but also utilizes my FancyApp, she could do so:

module JanesProgram
  TomsLib  = lib('tomslib', '1.0')
  FancyApp = lib('fancyapp')  # use newest available
  ...

How does this work? When you call lib(), Bezel looks for the package in the current gem paths (and, in the future, Roll ledger) then it reads the primary package file (eg. tomslib.rb) fro the package and evals it into an anonymous module.

This has a some important effects on how you write your Ruby programs:

1. Any reference to core/standard libraries must be referenced via

   prefix (eg. ::Enumerable).

2. Core extensions are not version controlled. So avoid them when possible, or depend on highly stable standardized bases such as Facets and ActiveSupport.

3. Since Bezel is a completely different alternative to Ruby’s normal load system, your program will require Bezel be installed by your users.

Despite these necessary practices for its use, Bezel is highly advantageous to the developers and end-users alike in that it puts an end to the dreaded Dependency Hell.

TODO: Consider how best to support alternate loadpaths beyond ‘lib/’.