diff --git a/lib/ruby/ruby_compiler.rb b/lib/ruby/ruby_compiler.rb
index 37d20e26..11f652dd 100644
--- a/lib/ruby/ruby_compiler.rb
+++ b/lib/ruby/ruby_compiler.rb
@@ -14,6 +14,7 @@ module Ruby
       @node.to_s[0 ... 200]
     end
   end
+
   # This RubyCompiler compiles incoming ruby (string) into a typed
   # version of theast, with the help of the parser gem.
   # The parser outputs an abstract ast (nodes)
diff --git a/test/mains/README.md b/test/mains/README.md
index e46d3433..ff11d89a 100644
--- a/test/mains/README.md
+++ b/test/mains/README.md
@@ -1,14 +1,18 @@
 # Mains testing
 
-Test methods by their output and exit codes (return, since it is the main). 
+Test methods by their output and exit codes (return, since it is the main).
 
-There are only two tests here, one for interpreter, one for arm.
+There are only two tests here (lus one, see below), one for interpreter, one for arm.
 Both run the same tests. The actual ruby code that is run is in the source dir.
 Test methods are generated, one for each source file.
 
+## Files
+
 File names follow [name,stdout,exitcode] joined by _ pattern. Stdout may be left blank,
 but exit code must be supplied.
 
+## Arm
+
 Obviously the arm tests need an arm platform. This may be defined by ARM_HOST,
 eg for simulated ARM_HOST=localhost
 
@@ -16,3 +20,11 @@ Also port and user may be specified with ARM_PORT and ARM_USER , they default to
 2222 and pi if left blank.
 SSH keys must be set up so no passwords are required (and the users private key may
   not be password protected)
+
+## Developing
+
+Since the Framework always runs all tests, it is a little cumbersome for developing
+a single new test. Since all get run and it is slow.
+
+To develop the next test, one can edit test_new.rb . Once it runs on the interpreter,
+move the changes to a source file and revert test_new changes.
diff --git a/test/mains/source/add__4.rb b/test/mains/source/add__4.rb
index 0b0fb26a..b9e9e66e 100644
--- a/test/mains/source/add__4.rb
+++ b/test/mains/source/add__4.rb
@@ -1,5 +1,5 @@
 class Space
   def main(arg)
-    return 2 +2
+    return 2 + 2
   end
 end
diff --git a/test/mains/source/simple-class__4.rb b/test/mains/source/simple-class__4.rb
new file mode 100644
index 00000000..edef374e
--- /dev/null
+++ b/test/mains/source/simple-class__4.rb
@@ -0,0 +1,8 @@
+class Space
+  def self.simple
+    return 2 + 2
+  end
+  def main(arg)
+    return Space.simple
+  end
+end
diff --git a/test/mains/test_new.rb b/test/mains/test_new.rb
index 74989e57..63e48bca 100644
--- a/test/mains/test_new.rb
+++ b/test/mains/test_new.rb
@@ -1,17 +1,40 @@
 require_relative 'helper'
 
+# The "New" in this TestNew means that this is the place to develop a new test for the mains dir.
+# If you just pop a file in the source directory, all tests will run. (Well, actually guard does
+#  not pick up changes on that source dir yet, but when you run it manually, all tests run)
+# This is fine if all is fine. But if all is fine, you are not developing, just playing.
+# So when you really need to itereate this editing this gives guard auto-run and just of that
+# one test that you work on.
+
+# After getting the test to run, copy paste the whole code into a file in source and revert
+# changes on this file.
+
 module Mains
   class TestNew < MiniTest::Test
     include Risc::Ticker
 
+    def whole_input
+      <<-eos
+        class Space
+          def self.simple
+            return 2 + 2
+          end
+          def main(arg)
+            return Space.simple
+          end
+        end
+      eos
+    end
+
     def setup
-      @string_input = as_main("a = 37 ; while(a>0) ; a = a - 1 ; end ; return a")
+      @string_input = whole_input
       super
     end
-    def test_chain # max 1011 iterations on 1014 integers (1024 - 10 reserve)
+    def test_chain
       run_all
       assert_equal ::Integer , get_return.class , " "
-      assert_equal 0 , get_return , " "
+      assert_equal 4 , get_return , " "
     end
 
   end