I noticed the missing line of code while chasing down an Unexpected byte found when reading an object: 50" exception that is thrown during a clustering test.

from irc:

<dmlloyd> ah well. I still think it's a bug. But tbh unexpected byte errors are most likely due to user marshallers
<dmlloyd> ClassTable and ObjectTable implementations are able to bypass the normal protections (the benefit is speed, the cost is... this)

For my failing clustering test case, adding the missing code to the ID_CLEAR_INSTANCE_CACHE case doesn't help, so I'll look elsewhere for the marshalling error.

In case it helps, 50 is a ID_START_BLOCK_SMALL.