IMHO, the svn log and the bug tracking system are not solely directed to library users.
Developers too might like to be able to follow why and how code changes.
I understand your concern, but "historically"
MATH-1010 was not meant to fix something, just to move code from one place to another; I fortuitously uncovered the bug because the method was intended to do more than is actually needed in "RandomDataGenerator" ("user" behaviour was indeed fine but "developer" code was misleading).
In order to not frighten users, I can add a comment in the "changes.xml" file to that effect; i.e. for
MATH-1020, something like: This bug does not affect applications using a previous version of Commons Math. For this issue, it should be obvious since the method "shuffle" was private.