"Failing to revert empty account deletions when the transaction causing the deletions of empty accounts ended with an out-of-gas exception" caused issues with consensus on November 24 2016.
Usually, a transaction provides a comfortable cushion of gas so that it doesn't reach an out-of-gas exception. For example, if one thinks their transaction might consume 500K gas, specifying 1M gas for it seems reasonable since all unused gas will be refunded.
Testing a transaction for Out of Gas (OOG) must also have been performed (because transactions either complete or OOG), so what are the details?
How did the "state clearing" transactions run OOG?
Were there several primary ways that they ran OOG? If so, how can they be classified so that the community keeps them in mind for their own testing?
