In order to simplify maintenance and allow us to eventually test the same thing on real browsers, either via puppeteer, electron, or actual browser, I suggest we decouple the benchmark from the rest of the table creation.
In this MR:
utilities are retrieved once
no way to deal directly with mutations (exposed count() as we did previously)
all foreigners dependencies are meomized once
Waiting for an opinion on this, before implementing puppeteer as extra benchmark.
In order to simplify maintenance and allow us to eventually test the same thing on real browsers, either via puppeteer, electron, or actual browser, I suggest we decouple the benchmark from the rest of the table creation.
In this MR:
count()as we did previously)Waiting for an opinion on this, before implementing puppeteer as extra benchmark.