adds the concern aproach and explanation

test/parser/README.markdown

@@ -14,20 +14,6 @@ Each test must thus specify (as instance variables):
 - the parse output
 - the transform output
 
-
-For any functionality that we want to work (ie test), there are actually three tests, with the _same_ name
-One in each of the parser/transform/ast classes
-Parser test that the parser parses and thet the output is correct. Rules are named and and boil down to 
-       hashes and arrays with lots of symbols for the names the rules (actually the reults) were given
-Transform test really just test the tranformation. They basically take the output of the parse
-        and check that correct Ast classes are produced
-Ast   tests both steps in one. Ie string input to ast classes output
-
-All threee classes are layed out quite similarly in that they use a check method and 
-each test assigns @input and @expected which the check methods then checks
-The check methods have a pust in it (to be left) which is very handy for checking
-also the output of parser.check can actually be used as the input of transform
-
-Repeat: For every test in parser, there should be one in transform and ast
-                                 The test in transform should use the output of parser as input
-                                 The test in ast should expect the same result as transform
+Test are grouped by functionality into cases and define methods parse_*
+Test cases must include ParserHelper, which includes the magic to write the 3 test methods for each 
+parse method. See test_basic ofr easy example.