mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-30 04:41:49 +00:00
5e3d5c5f6e
Summary: Since we have enough memory to hold all primary keys loaded from spatial index, it is better if we first load all of them (store them in unordered_set for deduplication) and then query on primary key column family one by one. We need to dedup all IDs, so we'll end up storing all of them in memory even with the current approach. Test Plan: ./spatial_db_test is happy Reviewers: yinwang Reviewed By: yinwang Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D20949 |
||
---|---|---|
.. | ||
spatial_db.cc | ||
spatial_db_test.cc | ||
utils.h |