Closed Centril closed 3 months ago
benchmarks please
Benchmark results appear to be within noise. The noise is admittedly quite large, and I don't think we have any benchmarks that hit the subscription manager, but what's important is that this isn't a huge regression to query performance. The actual improvements will come later, when we combine this with #840 , do follow-ups and optimize MemTable
.
I think this was potentially merged prematurely @gefjon and @Centril. This is going to cause a lot of churn for Mario and Joshua as they are working on this code right now. Given that this doesn't directly improve performance, adding additional burden on them to rebase doesn't seem to make sense to me.
I think I actually need to revert this one to let Mario and Joshua finish their work first.
Description of Changes
The new definition is:
The PR also removes other unnecessary clones, including some notable ones, e.g.,
MemTable::next
.API and ABI breaking changes
None
Expected complexity level and risk
2