I see a strange thing in my rake script today. I have two Rake tasks under the different namespaces like following:
path = "/home/tomcat/tomcat"
namespace :stage do
path = "/home/tomcat/stage-tomcat"
desc "Deploys a java application to stage tomcat"
task :java_deploy do
puts path # stage:java_deploy should print /home/tomcat/stage-tomcat
end
end
namespace :production do
path = "/home/tomcat/production-tomcat"
desc "Deploys a java application to production tomcat"
task :java_deploy do
puts path # production:java_deploy should print /home/tomcat/production-tomcat
end
end
When I run: rake stage:java_deploy it prints
/home/tomcat/production-tomcat
I was expecting /home/tomcat/stage-tomcat. If I remove the very first line path = "/home/tomcat/tomcat" from the rake file, it works as expected.
Any idea why this kolavari? :)
Thanks in advance!!
This is not specific to Rake, it is just a consequence of lexical scope and the way Ruby handles local variables, declaring them on first use.
First you assign a value to path:
path = "/home/tomcat/tomcat"
Then you create the stage namespace and reassign the variable:
path = "/home/tomcat/stage-tomcat"
Note that this line is executed whatever task you specify, as it is not in any tasks.
Next you create the java_deploy task, but it doesn’t get run yet. This task refers to the path variable, but when the task is called the value of it might have changed.
Later, when defining the production namespace this variable gets reassigned again. Importantly this is still the same variable:
path = "/home/tomcat/production-tomcat"
When the task is actually run, it refers to the path variable and the value of this variable is the latest value assigned to it, which is /home/tomcat/production-tomcat.
When you remove the first assignment to path, then the variable doesn’t exist in the top level. This means that when you assign to path in each namespace definition you are declaring a new (and separate) local variable in each case.