I suspect the problem of #27 occurs because no row insertion dependency is considered in the flush() method. That could be fairly complex to resolve, it seems to me.
Not using write buffering at all, which this commit allows, certainly slows down the process a bit, but it does allow the subset extraction to work for tangled database relationships. Better slow and complete than broken :).
By the way, thanks for writing this! It's very helpful!
This fixes the issue reported in #27
I suspect the problem of #27 occurs because no row insertion dependency is considered in the
flush()
method. That could be fairly complex to resolve, it seems to me.Not using write buffering at all, which this commit allows, certainly slows down the process a bit, but it does allow the subset extraction to work for tangled database relationships. Better slow and complete than broken :).
By the way, thanks for writing this! It's very helpful!