
The previous use of require caused File.join on several occasions to calculate different paths to the same library, depending on which __FILE__ the library was being calculated as relative to; e.g. /some/path/prefix/spec/one/bar.rb would do the equivalent of: require '/some/path/prefix/spec/one/../../libraries/foo/mylib.rb' and /some/path/prefix/spec/two/baz.rb would do the equivalent of: require '/some/path/prefix/spec/two/../../libraries/foo/mylib.rb' This would result in mylib.rb being loaded multiple times, causing warnings from constants being redefined, and worse, multiple objects representing the same class hierarchy (@@foo) variables. The latter actually broke the @@subclasses registration mechanism in Pacemaker::CIBObject. By switching to File.expand_path, we ensure we always refer to each library using a single absolute path, which means Ruby's require mechanism works as it should, only loading the code the first time round.
11 lines
233 B
Ruby
11 lines
233 B
Ruby
require File.expand_path('cib_object', File.dirname(__FILE__))
|
|
|
|
module Pacemaker
|
|
class Constraint < Pacemaker::CIBObject
|
|
def self.description
|
|
type = self.to_s.split('::').last
|
|
"#{type} constraint"
|
|
end
|
|
end
|
|
end
|