Recently on the Pharo list there have been some complains of loosing code when filing out change sets.

In 3.10 changeset code was augmented to show which package catagory a method fell into. This is essential if coders are to know how MC and the changeset will interact.

Those changes mucked with some of the fundamental ivar assumptions. And while I thought I caught them all I know of at least two that slipped by.

To safely work on the code it would be nice to have regression tests in place. For that I would like to know if anyone thought of the use cases that need to be tested or any actual unit test code that might exist.

What I am hoping for is for each type of thing that could be filed out, a way to test that changes to the code do not change the final fileout behavior. Then I can confidently do some refactoring.

Yours in curiosity and service, --Jerome Peace