Refactor construction to avoid new

[jsumners]

I'm not sure if you guys want this. I just wanted to see how Pino would perform if it didn't rely on the new construct. I am attaching single runs of the full benchmark suite from the current master and this refactor branch. It would be nice to know how the runs compare on some other systems. I'll let you be the judge of the results.

There are a lot of changes here. You may want to simply view the full file instead of the diff. I started the work from the issue-209 branch in PR #210.

One nice change here, in my opinion, is that we don't have to expose the _Pino property for things like pino-multi-stream. Those sort of modules can simply get a new instance out of the factory function like normal and do true JavaScript "inheritance".

refactor.bench.txt
master.bench.txt

[mcollina]

benchmarks are good and just a bit better, just double check but it seems the 'object' one is slightly slower than master. Maybe it was just a bad run, but better check.

If that's ok, 👍 for me.

[jsumners]

I think it's just variations in CPU scheduling. I am attaching the results of three consecutive runs of bench-object on both master and refactor branches. The second run of each shows this. And in each of the three runs the PinoUnsafeExtremeObj results are nearly identical.

bench-object.refactor.txt
bench-object.master.txt

[davidmarkclements]

oh very nice - I refactored to use constructor because of perf benefits over closure

but if we can use Object.create (which is fine to use now since we're not trying to support v. old JS) instead of new then all the better 👍

[davidmarkclements]

LGTM

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.