Use packed cached hashes for faster table mutations - #962

Draft

ezrosent wants to merge 12 commits into

codex/demand-driven-join-schedulingfrom

ezr-better-rebuild

Commits on Aug 10, 2026

Store predicted rows contiguously
ezrosent
committed
fba3c26
Reserve fresh IDs per execution state
ezrosent
committed
100f879
Explain counter reservation lifecycle
ezrosent
committed
2eeaaec
Build column indexes from parallel sorted runs
ezrosent
committed
70dc650
Use coarse partitions for table rebuilds
ezrosent
committed
144cb33
Benchmark production table parallelism
ezrosent
committed
5c14f69
Optimize parallel table deletion
ezrosent
committed
107ef48
Benchmark large parallel table deletion
ezrosent
committed
641f425
Partition table mutations by cached hashes
ezrosent
committed
d83bd18
Add math scalability benchmark harness
ezrosent
committed
808db86
Improve parallel union-find and join scalability
ezrosent
committed
ddc35f0
Use packed cached hashes for table mutations
ezrosent
committed
5ec34b7