BuilderIO / framework-benchmarks

Test each framework for it's performance cost
473 stars 22 forks source link

Make react-dom ssr section more accurate #15

Closed Jarred-Sumner closed 2 years ago

Jarred-Sumner commented 2 years ago
Results using the same react-dom/server:

None of the runtimes are using the same build of react-dom/server, so this line is incorrect.

Node is using react-dom/server.node: https://github.com/BuilderIO/framework-benchmarks/blob/2fb93ecf448737ab08b5e2efa481d54a709fce97/frameworks/react-ssr-node/http.jsx#L4

Deno is using react-dom/server.browser: https://github.com/BuilderIO/framework-benchmarks/blob/2fb93ecf448737ab08b5e2efa481d54a709fce97/frameworks/react-ssr-deno/http.jsx#L5

Bun is using what will become react-dom/server.bun: https://github.com/BuilderIO/framework-benchmarks/blob/2fb93ecf448737ab08b5e2efa481d54a709fce97/frameworks/react-ssr-bun/http.js#L5

In real-world usage, it's not clear why someone would choose a build intended for a different host, so it's not particularly useful for benchmarking.

Also, we can switch bun's import.meta.require to be a regular import statement. There was a bug little while ago that has since been fixed

Are the tables manually updated? I just moved the bun row from the bottom table to the top table and removed the bottom table

steve8708 commented 2 years ago

Thanks @Jarred-Sumner, totally works for me. And yeah completely manual for now until we setup a proper automated way and remote (e.g. CI) env to run the results