michaeljbishop / rake-subproject Goto Github PK

The super-project will catch Exceptions that the sub-project raises as they are carried across the pipe, but they shoudl be reported in the cause field of the proxy exception raised in the super-project execution.

Currently, the subproject call expects a directory. It could also be a path to a Rakefile.

In other words, prevent this:

subproject 'foo'
subproject 'foo/bar' # fail here. it's up to `foo` to include `bar`

With the server in place, sending a SIGKILL to the main process should send the same to the sub-process and kill it.

Rake allows a -j parameter which specifies the maximum number of jobs. This number should work not just for the super-project, but the sub-projects as well. To make matters more difficult, if the super-project can fill all the job-spots, then the sub-project shouldn't get any.

There should be communication between the super and sub-project instances as to how many tasks are being executed and if any more can without exceeding that limit as established in the super-project execution.

Right now there is no special code that checks that the subproject call is passed a valid directory or Rakefile.

Add some more parameters to the subproject call that would allow more configuration of the subproject. In particular, there should be some way of setting up the environment variables.

Issues to address:

1. Should the super-project pass along all of the Rake ENV that it processed at the command-line?
2. What should the default be?

As long as the super-project can reference tasks created in the sub-project, allow the sub-project tasks to have prerequisites attached to them in the super-project Rakefile.

So, to be clear, here are some tasks in in the subproject, located at foo.

task :hello do
  p "HELLO"
end
task :world => :hello do
  p "WORLD"
end
task :cruel do
  p "CRUEL"
end

And here is the super-project:

subproject 'foo'

task 'foo:world' => ['foo:cruel', 'period'] do
  p "AGAIN"
end
task 'period'  do
  p "."
end

Now, when executing the super project:

$ rake foo:world
HELLO
CRUEL
.
WORLD
AGAIN

The sub-project prerequisites are all executed before the super-task prerequisites.

To reproduce, make a Rakefile with a task that depends on a Rakefile 2 subprojects away. Have that leaf task raise an expection. The top-level project will simply sit forever waiting for the task to complete.

As long as the super-project can reference tasks created in the sub-project, allow those tasks to have execution blocks attached to them in the super-project Rakefile.

So, to be clear, here's a task in in the subproject, located at foo.

task :hello do
  p "HELLO"
end

And here is the super-project:

subproject 'foo'

task 'foo:hello' do
  p "WORLD"
end

Now, when executing the super project:

$ rake foo:hello
HELLO
WORLD

The sub-project blocks are all executed before the super-project blocks.

And undocument the private ones.