Bugfix/48

[Pho3niX90]

closes #48

[pkg-pr-new]

Open in Stackblitz

npm i https://pkg.pr.new/dimdenGD/ultimate-express@49

commit: b938cc4

[dimdenGD]

There's no 20.70.0 uWS.js, you're confusing it with C++ uWS

[Pho3niX90]

you are right. I was looking at the C++ version.

[Pho3niX90]

@dimdenGD I am finished with this now if you want to review.
I have added a benchmark workflow as well, super basic, and can post to the PR, I think you just need to enable it in your settings. Would give a nice indication if a PR is slowing the code down etc.

[dimdenGD]

Why do {...new NullObject()}? It's like creating a fast object and then creating a slow one and putting it inside. Would be faster to just have {} at this point

[Pho3niX90]

Why do {...new NullObject()}? It's like creating a fast object and then creating a slow one and putting it inside. Would be faster to just have {} at this point

My initial thoughts. But ran some benchmarks, and seems it's faster like this. See if you get the same results from it?

[dimdenGD]

I don't see how it's possible for it to be faster that way. Microbenchmarks are often deceptive

[Pho3niX90]

So these are the results (tests/sample/object_creation.js):

Min is not the lowest, but more consistent than the spread.

But I am fine with either. Should I change it back to {}

[dimdenGD]

You should probably test some reads+writes too. Also there's no point in using these spreads for internal stuff only used by library, only for public things like req.query and req.params