- Notifications
You must be signed in to change notification settings - Fork5
Commit4e7d10c
committed
Comments in IndexBuildHeapScan describe the indexing of recently-dead
tuples as needed "to keep VACUUM from complaining", but actually there isa more compelling reason to do it: failure to do so violates MVCC semantics.This is because a pre-existing serializable transaction might try to usethe index after we finish (re)building it, and it might fail to find tuplesit should be able to see. We got this mostly right, but not in the caseof partial indexes: the code mistakenly discarded recently-dead tuples forpartial indexes. Fix that, and adjust the comments.1 parent0a20207 commit4e7d10c
1 file changed
+5
-6
lines changedLines changed: 5 additions & 6 deletions
Original file line number | Diff line number | Diff line change | |
---|---|---|---|
| |||
8 | 8 |
| |
9 | 9 |
| |
10 | 10 |
| |
11 |
| - | |
| 11 | + | |
12 | 12 |
| |
13 | 13 |
| |
14 | 14 |
| |
| |||
1473 | 1473 |
| |
1474 | 1474 |
| |
1475 | 1475 |
| |
1476 |
| - | |
| 1476 | + | |
| 1477 | + | |
| 1478 | + | |
1477 | 1479 |
| |
1478 | 1480 |
| |
1479 | 1481 |
| |
| |||
1541 | 1543 |
| |
1542 | 1544 |
| |
1543 | 1545 |
| |
1544 |
| - | |
1545 |
| - | |
| 1546 | + | |
1546 | 1547 |
| |
1547 | 1548 |
| |
1548 | 1549 |
| |
1549 |
| - | |
1550 |
| - | |
1551 | 1550 |
| |
1552 | 1551 |
| |
1553 | 1552 |
| |
|
0 commit comments
Comments
(0)