Object.assign is built-in and has some performance advantage-
over the polyfill.

Do you have some benchmarks to back that up? That wasn't the case, as of #4593.

Also see this similar PR that was submitted just the other week.

@cjihrig with large sample I can consistently see over 10% performance gain using Object.assign.

Here is the test script and the result.

@cjihrig I just made a couple of minor changes to the script, however the results remain around the same.

I will close this pr in favour of any one of the similar earlier prs referenced here.

However, would like to hear if anyone has any thoughts on the benchmark findings using the attached script.

However, would like to hear if anyone has any thoughts on the benchmark findings using the attached script.

I think it was the nature of the data you were operating on. I switched to util._extend(extend, process.env); and Object.assign(extend, process.env); and saw a much bigger difference, in favor of util._extend().

$ ./node cji.js 
_extend took:: 186755.394ms
oassign took:: 319979.233ms
_extend took:: 200674.158ms
oassign took:: 346792.866ms
^C

It might be good to have a benchmark in the repo for this, since it keeps coming up.

@cjihrig this is interesting. With your process.env approach, except the first iteration of the while loop, the source and destination objects are always similar structured objects. And in this case _extend() looks to be much faster. Seems optimizer (v8?) is able to detect this.
However, when the source and destination objects are disjoint structured, i.e, every property of the src needs to be copied to the dest, then Object.assign() looks to be slightly faster.

@suryagh can I suggest opening a PR to create a benchmark for this?

cc: @nodejs/benchmarking

@cjihrig will do.

#7255