Completely hide dataset when data is empty or null

[BelleNottelling]

Feature Proposal

I think it would make a lot of sense for a dataset to not even show an empty box in the legend when the data is empty or null. Now, I'm really not how many people experience this as an issue, it is very possible that this is just another example of me not being particularly great at programming and doing something dumb as a result

Feature Use Case

So, I'm building a little web app to help me analyze ECU logs for tuning purposes.
The problem? There is no way for me to know how many datasets, what the dataset is of, or what the dataset is called without actually reading it from the log.

My code has to read and "map" each sensor name for each log as there's no guarantee that the data will be in the same format between logs. I basically just have a bunch of empty datasets that get populated, but when not all of them are needed I'm left with a bunch of empty boxes for data that doesn't exist. The more data I want to be able to support, the worse it gets.

So I'm not really sure if it would be possible to dynamically create each dataset as I need rather than having a ton of empty dataset that I just fill in, or if it would be possible to simply automatically hide the empty datasets.
The source code for my project can be found here
To visually see what I'm talking about you can view it on my website and load the log attached to this ticket
romraiderlog_log.t_20210405_210451.csv.zip

Now, this is a "feature request" in the manner of where I think hiding an empty dataset makes sense, but it's also me wondering if there's a better way for me to go about this. I appreciate any input, thanks!

[kurkle]

If you pre-create the datasets for having some predefined styles, then you should just define the styles beforehand as an array and map that to the datasets you create when reading the log. And only create as many datasets as there are.

Or, if you prefer less work, you could just do something like

const datasets = createMyDatasets(); // your existing creation code, did not take a look

const chart = new Chart('id', {
  data: {
    datasets: datasets.filter(ds => ds.label !== '')
  },
  ...
});

[BelleNottelling]

Hi, sorry for not using the discussions feature, this is actually the first repository I've seen using it so I didn't check!
I previously tried using the legend filter to remove those boxes, but obviously did it wrong since I ended up without any data. The way you suggested seems to work great. Thank you!

This isn't strictly related to Chart.js, but the way I handle my datasets is pretty poor.
It's defined like this, and in a separate function I use counting and an eval statement to set these numbers. I know this is bad for performance and for security, and it means I need a bunch of empty datasets for this to work right.

const data1 = {
  labels: timelabels,
  datasets: [{
    label: GenericSensors._1.SensorName,
    data: GenericSensors._1.Values,
    fill: false,
    borderColor: '#a6cee3',
    tension: 0.1,
	yAxisID: 'left-y-axis',
	hidden: true
  },

and

var GenericSensors = {
  _1: {
    HeaderIndex : -1,
    SensorName : "",
    Values :[]
  },

and

	    sensorcnt++;
	    var sensname = "'" + item + "'";
	    eval("GenericSensors._" + sensorcnt + ".Values.push(" + sensname + ")");//More evil thing. This one may actually be bad

Can you point me in the right direction for creating the datasets in a non-predefined way?
Again, this isn't really related to Chart.js so I'm not expecting anything, but I'd really appreciate it.
Thanks again!