I'm trying to decide if I want to remove the last vestiges of record numbers in row-store databases: we could make record numbers work in append-only row-stores, but that's about it, inserts aren't possible, at least not when any scheme I've seen. Thoughts? Do you see any reason to support some limited form of record number access in row-stores?
- is related to
-
WT-543 LSM core dump in the checkpoint code
- Closed
-
WT-581 test/format crash
- Closed
-
WT-680 test/format failure: LSM/compaction deadlock
- Closed
-
WT-682 wtperf failure: accessing a freed page
- Closed
-
WT-713 wtperf parallel-pop-lsm failure
- Closed
-
WT-961 Checkpoints can lose information
- Closed
-
WT-976 format checksum error
- Closed
-
WT-1016 Transaction visibility assertion firing
- Closed
-
WT-1017 voxer 10k - reconciliation illegally skipped an update
- Closed
-
WT-1245 Test format stress failure - strange LSM stack
- Closed
-
WT-1280 test/format eviction segfault in hazard_exclusive
- Closed
-
WT-1296 Heap use after free
- Closed
-
WT-1306 LSM heap-use-after-free test/format
- Closed
-
WT-1307 LSM read checksum panic
- Closed
-
WT-1736 test/format sanitizer memory leak
- Closed
- related to
-
WT-1 placeholder WT-1
- Closed
-
WT-2 What does metadata look like?
- Closed
-
WT-3 What file formats are required?
- Closed