I'm looking for a PHP library or process for comparing large hierarchical data sets and giving some form of human readable analysis. The purpose is for analyzing regressions in the process that created them. My current method of testing in PHPUnit is:
Iterate an existing data set (saved as serialized data or loaded directly from an API), and build an equivalent data set using the application API methods.
Compare the two datasets by basically MD5'ing various segments and using assertEquals(). If it fails then manually diff the output to actually learn things.
It works, but its fragile for several reasons eg. the addition of a new field will always result in a failure. It's part of an automated workflow so I need a way to exclude these kind of false positives and generally analyze and report on it better.
