hashbrown 0.15 was released this month, which changed its hasher from ahash to the faster foldhash. PR 445 updates egglog, but index_set keeps its own older version of hashbrown—hence this PR. This PR updates hashbrown and switches to foldhash.
The egglog benchmarks for the other PR only show regressions for what Saul says are unreliable tests with high variability. Testing updating symbol_table through those benchmarks doesn't seem to show any difference from a version which doesn't change symbol_table.
I wasn't able to figure out how to run the Criterion benchmarks in symbol_table—the typical invocations weren't doing the trick—, but that should give better insight.
hashbrown
0.15 was released this month, which changed its hasher fromahash
to the fasterfoldhash
. PR 445 updates egglog, butindex_set
keeps its own older version of hashbrown—hence this PR. This PR updates hashbrown and switches to foldhash.The egglog benchmarks for the other PR only show regressions for what Saul says are unreliable tests with high variability. Testing updating symbol_table through those benchmarks doesn't seem to show any difference from a version which doesn't change symbol_table.
I wasn't able to figure out how to run the Criterion benchmarks in symbol_table—the typical invocations weren't doing the trick—, but that should give better insight.