Closed leeoniya closed 6 months ago
done.
this actually lost some perf. i was able to get some of it back, but not all. net loss it maybe up to 10%, depending on corpus.
in Node/V8 it's a small improvement in mem, but in Bun/JSC it's a huge mem reduction (>50%).
i think it will reduce mem use a lot if we just
.slice()
from start to end and do a.replaceAll()
to remove the escaped commas or quotes. not sure this can work for incremental parsing but probably. think this will increase overall perf as well since mem alloc / GC is the slowest thing in JS.