I am trying to work on this issue, but I need more details when NaN is passed on the line chart. For the Bar chart type, I set the height and width of the bar with the NaN value to 0, so the chart does not render that specific bar. However, since line charts are continuous, I'm not sure at which point I should break the line and restart it. @jgclark It would be helpful if you could show me how other graphing packages deal with this issue.

Thank you for your work on this.
gnuplot handles this by disconnecting the line where required, so in your example you'd just get a dot on data point 1 and then nothing until data point 3 where the line would start and continue on. You can actually specify what character to use (or blank space) to indicate a missing point: e.g. set datafile missing '?' or set datafile missing 'NaN'. Fuller details: https://gnuplot.sourceforge.net/docs_4.2/node172.html.
For example:

Google Chart API by default leaves a similar break, according to its documentation https://developers.google.com/chart/interactive/docs/gallery/linechart?csw=1#Configuration_Options under InterpolateNulls.

Yes thank you for your work on this.
In my usecase I want to draw a line for i.e. x=0 y=100 to x=100 y=0. I have other lines on that chart which have more datapoints between x 0 and 100. Be able to draw that line without supplying the values inbetween would be nice.

Sorry for taking a while to work on this; I made changes so that the passed-in NaN values break the line before and after the missing point. I wonder if we should include options so that users can customize how NaN values are represented. One possibility is to assign a different theming variable (maybe with a dashed line or lighter color as shown in https://observablehq.com/@d3/line-chart-missing-data/2?intent=fork)

This would probably be a separate issue after this one has been resolved.

Thanks for your work on this.
I think I would start a separate issue for the "customize how NaN values are represented. One possibility is to assign a different theming variable (maybe with a dashed line or lighter color" idea. Some will want this ability for certain datasets, but for others it would be misleading, so should default off.

спасибо большое