Serial Log debugging

[Christopher Powers]

All,

Bug#39696, "Assertion in Table.cpp (dup->state == recDeleted) during 
falcon_chill_thaw" is finally fixed. There were two very subtle bugs in 
Transaction::rollbackSavepoint() that occasionally prevented chilled 
records from being fully committed. One of these bugs would've likely 
manifested as a lost update had it occurred outside of chill/thaw, but 
it's hard to say.

Now falcon_chill_thaw triggers the familiar assertion in 
SerialLogRecord::nextRecord():

    ASSERT(*input >= (LOW_BYTE_FLAG | srlEnd) && *input < (LOW_BYTE_FLAG 
| srlMax));

This happens every time within seconds. It appears that the length 
counts are off somewhere, but where?

Can any of you who've debugged this sort of thing (Jim, Ann, Olav?) 
suggest an efficient approach?

Chris