Here is some sample input.

File1
-----
X  Y  Category1 Category2 Value1 Value2 Value3 Result 
2  -9 1.0       2.0       1.1e3  1.234  -0.003 PASS   

File2
-----
X  Y  Category2 Category1 Value3 Value1 Value2        
2  -9 1         1         -0.003 1.1e3  1.2345 FAIL
[download]

The main points to note are:
1. The columns are not necessarily in the same order (the main reason I started this in the first place).
2. Data is a mixture of strings and numbers of differing precisions (but all base10).

To take account of the differences in column order, I rebuild the data using a hash keyed by the XY coordinate (first 2 columns) and using the actual column name e.g

...other code omitted...
$data1{"$x,$y}{$colnames[$colnum]} = $linedata[$colnum];
[download]

Output would be something like:
X=2 Y=9 Category2
X=2 Y=9 Value2
X=2 Y=9 Result


i.e. a list of the column names for which the data did not match between the two files, allowing for strings an numerical values, and accuracy to a certain precision (using the $eps approach detailed elsewhere in this thread)