add visibility state and toggling of datasets

[ennerf]

This is an outline for what the toggle feature could look like. It's not ready for merging. Below are some questions I've run into so far:

1. What's the best way to omitt the hidden datasets in the axis range computation? Ideally something that could work on the Chart baseclass?

2. Is there a good way to deal with access to the dataset, e.g., EditDataSet without supporting it in every plugin?

[RalphSteinhagen]

Very nice PR and good addition to chart-fx! 👍

[ennerf]

I added the suggested changes and ran through all samples. The IMO noteworthy ones are:

EditDataSetSample
Hiding a dataset does not remove the point selection. Without modifications it is possible to select both datasets, hide one, and then apply operations on hidden points resulting in indexing errors. I changed it such that points remain selected, but that any modifying operations are ignored.

MetaDataRendererSample
hiding random walk also hides the gaussy dataset. Maybe this is actually be the desired behavior?

[RalphSteinhagen]

Hi @ennerf looks good.

It doesn't compile in the CI/CD due to JavaFX dependencies in non-UI modules (new BooleanProperty used for visibility state) but that IMO could be easily replaced by a simple boolean isVisible state and using the existing event notification mechanisms, e.g. in AbstractDataSet

    public boolean isVisible() { return isVisible; };

    public void setVisible(boolean visible) {
        if (visible != isVisible) {
            isVisible = visible;
            fireInvalidated(new UpdatedMetaDataEvent(this, "changed visibility"));
        }
    }

With these changes it should compile/work as expected. Please have a look (e.g. merge) also at the code-formatter (Restyled.io bot PR) and static-code analysis checks.

After that this could be merged. Thanks for this very nice PR! 👍

[wirew0rm]

Sorry, I kind of missed the updates in this PR.

MetaDataRendererSample hiding random walk also hides the gaussy dataset. Maybe this is actually be the desired behavior?

This is actually a bug in the sample I also discovered when I investigated the 11.2.6 axis regression. The gaussy dataset implements incorrect range information by overwriting AbstractTestFunction's get() function but not updating the range. So when you toggle visibiltiy of the other datasets, the axis range changes to something outside of the gauss functions x-value range. You probably can get it to show up again by manually zooming. I have a partial fix for that here, so don't worry about this one.

[codecov]

Codecov Report

Merging #481 (f4b7b7a) into master (89a5ad3) will decrease coverage by 0.00%.
The diff coverage is 43.63%.

@@             Coverage Diff              @@
##             master     #481      +/-   ##
============================================
- Coverage     53.63%   53.62%   -0.01%     
- Complexity     7477     7489      +12     
============================================
  Files           389      389              
  Lines         40952    40989      +37     
  Branches       6604     6615      +11     
============================================
+ Hits          21963    21982      +19     
- Misses        17404    17417      +13     
- Partials       1585     1590       +5     

Impacted Files	Coverage Δ
...rc/main/java/de/gsi/chart/plugins/EditDataSet.java	23.21% <0.00%> (-0.19%)	⬇️
...gsi/chart/renderer/spi/ContourDataSetRenderer.java	76.04% <0.00%> (ø)
...a/de/gsi/chart/renderer/spi/HistogramRenderer.java	66.25% <0.00%> (-1.16%)	⬇️
...gsi/chart/renderer/spi/LabelledMarkerRenderer.java	83.87% <0.00%> (-1.10%)	⬇️
.../gsi/chart/renderer/spi/MountainRangeRenderer.java	0.00% <0.00%> (ø)
...e/gsi/chart/renderer/spi/ReducingLineRenderer.java	0.00% <0.00%> (ø)
...rt/renderer/spi/financial/CandleStickRenderer.java	88.60% <0.00%> (-0.64%)	⬇️
.../chart/renderer/spi/financial/HighLowRenderer.java	85.33% <0.00%> (-0.67%)	⬇️
...-dataset/src/main/java/de/gsi/dataset/DataSet.java	100.00% <ø> (ø)
...e/gsi/chart/renderer/spi/ErrorDataSetRenderer.java	79.11% <33.33%> (-0.37%)	⬇️
... and 9 more

---

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 89a5ad3...f4b7b7a. Read the comment docs.

[wirew0rm]

Very nice feature, I really like how it turned out!

[milo-gsi]

Thanks for providing that very helpful feature, Florian!
Just stumbled over it "accidentally" and I love it.