Problem: a lot of problems are created by tracking the importing progress on multiple tables. Querying and updating involves as such heavy operations and this is also the cause for inconsistencies. Solution: use a new table to track the progress of block's data indexing. By using a specific table, whose rows are deleted when there are no more pending operations on the relative block, we make joins quick and keep them to a minimum. This is also meant to simplify all inserts and updates and only perform deletes when strictly necessary.pull/2888/head
parent
2bd6f66e99
commit
036c16fb09