Restored old performance in ChartDataSet

[jjatie]

for #3166
Temporary fix for calcMinMax frequency.
Better implementation coming for Charts 4

[codecov-io]

Codecov Report

Merging #3216 into master will decrease coverage by 0.01%.
The diff coverage is 0%.

@@            Coverage Diff             @@
##           master    #3216      +/-   ##
==========================================
- Coverage   22.89%   22.87%   -0.02%     
==========================================
  Files         116      116              
  Lines       15609    15618       +9     
  Branches      272      272              
==========================================
  Hits         3573     3573              
- Misses      12000    12009       +9     
  Partials       36       36

Impacted Files	Coverage Δ
...s/Data/Implementations/Standard/ChartDataSet.swift	36.09% <0%> (-0.99%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 48bc049...937e8c2. Read the comment docs.

[liuxuan30]

hmm.. what PR is before this invert? I don't remember you particularly have a performance enhance PR？

[jjatie]

I don't know which one was the problem. I think it was when I moved to the didSet. I think it's important we have a fix for now, I have a good idea on how to improve it in 4.0.

[liuxuan30]

Weird I didn't catch this performance drop.

[liuxuan30]

wait a minute. I checked it's #3000

    @objc open var values: [ChartDataEntry]
    {
        get
        {
            return _values
        }
        set
        {
            _values = newValue
            notifyDataSetChanged()
        }
    }

I didn't see didSet or set has any difference

[liuxuan30]

eh, I kind of start to recall. didSet will be called for Array type every time an element is added in or removed (and replace), is that correct? That's the drop.

However for set setter, add/remove does not trigger the setter. Damn, I should remember this all the time.

[liuxuan30]

@jjatie do you think it's the right behavior for didSet on collection types? I will keep an eye on if swift evolution would address this. Kind of confusing.

[jjatie]

Yes, that’s the reason why. didSet is called whenever the underlying data mutates (excluding direct memory access). A computed variable doesn’t have underlying data, so it doesn’t make sense for it to be called on append/remove.

There’s a couple of major architectural changes I’m going to make over the next couple weeks. After that I’ll work in making performance better for 4.0

[liuxuan30]

@jjatie take some rest.