An example would be better than any explanation.

my $headers = {
    locations   => ['USA', 'UK'],
    departments => ['pers', 'dev','sales'],
    gender      => ['m', 'f']
};

__END__
The ultimate goal will be
                                
USA                                UK                                 
+   
---------------------------------- ---------------------------------
pers     dev     sales     USA     pers     dev     sales     UK  
-------- ------- --------- ------- -------- ------- --------- ------
m f pers m f dev m f sales m f USA m f pers m f dev m f sales m f UK
- - ---  - - --- - - ----- - - --- - - ---- - - --- - - ----- - - --

a b c    d e f   g h i     j k l   m n o    p q r   s t u     v w x
[download]

Take this schema, and see why I said that order is important and I take the variation only when the last one is changing.

For every location I list the departments; for each department a breakdown by gender (columns a b d e n o p q s t), followed by a department total (columns c f i o r u). Then there is a breakdown by location (columns g h v w) with its grand total (columns l x).

If I permute 'USA'-'m' before 'USA'-'pers'-'m' I am breaking the schema.

The algorithm I am dealing with is a little more complex than this, since it is taking information about where to find each item, and generating code to retrieve the relevant data from a database.

The permutations are needed to create the appropriate conditions for retrieving each column of data. All the conditions are merged together into a giant SQL statement that will fetch the data according to my instructions (SUM, COUNT, AVG).

More on this subject when I adjust nicely all the above into a module. :)

 _  _ _  _  
(_|| | |(_|><
 _|