Merged PRs
dolt
- 4821: Change dolt_diff_summary table function to return empty diffs for tab…
…les with pk set changes.- The 3 argument version returns an error if the table specified has a pk set change.
- The 2 argument verison returns empty diffs and returns a warning.
I considered approximating the diff for the pk set change, but it feels like an overreach for the customers need.
Fixes: #4818
- 4769: Add bats for column truncation on import
Associated GMS PR:
dolthub/go-mysql-server#1403
go-mysql-server
- 1408: hashlookups can hash byte slices
fix for: #4682 - 1407: adding tests for unique indexes
verifies this was fixed: dolthub/go-mysql-server#571
pulls aduplicate key update
test from dolt - 1406: in memory implementation of prefix indexes
Closed Issues
- 4818: SELECT * from dolt_diff_summary('commitA', 'commitB') should not abort if a primary key change is detected.
- 4474:
dolt migrate
fails with "migration failed cause: Incorrect datetime value" - 571: Unique indexes do not work in the in-memory implementation
Latency
Read Tests | MySQL | Dolt | Multiple |
---|---|---|---|
covering_index_scan | 1.96 | 2.71 | 1.4 |
groupby_scan | 12.3 | 16.71 | 1.4 |
index_join | 1.18 | 4.57 | 3.9 |
index_join_scan | 1.14 | 3.89 | 3.4 |
index_scan | 30.81 | 53.85 | 1.7 |
oltp_point_select | 0.15 | 0.47 | 3.1 |
oltp_read_only | 3.02 | 8.74 | 2.9 |
select_random_points | 0.3 | 0.75 | 2.5 |
select_random_ranges | 0.35 | 1.12 | 3.2 |
table_scan | 30.81 | 62.19 | 2.0 |
types_table_scan | 71.83 | 189.93 | 2.6 |
reads_mean_multiplier | 2.6 |
Write Tests | MySQL | Dolt | Multiple |
---|---|---|---|
bulk_insert | 0.001 | 0.001 | 1.0 |
oltp_delete_insert | 3.02 | 10.46 | 3.5 |
oltp_insert | 1.44 | 2.71 | 1.9 |
oltp_read_write | 5.18 | 16.71 | 3.2 |
oltp_update_index | 1.52 | 4.74 | 3.1 |
oltp_update_non_index | 1.47 | 5.0 | 3.4 |
oltp_write_only | 2.3 | 7.98 | 3.5 |
types_delete_insert | 3.07 | 12.75 | 4.2 |
writes_mean_multiplier | 3.0 |
Overall Mean Multiple | 2.7 |
---|