Fix ggplot2 update

[nikosbosse]

version 3.4.0 of ggplot2 deprecated the use of "size" in geom_line() and similar.
The problem is that I think in older version you can't use linewidth and it will error. So not sure when the best moment is to switch.
Also I think the tests generate lots of warnings due to the same issue in the ggdist pacakge. @Bisaloo do you have an idea about what the best thing forward would be?

[Bisaloo]

You should be fine if you set 3.4.0 as the minimum required version for ggplot2 in DESCRIPTION. This is what we have done for ggdist for example:

scoringutils/DESCRIPTION

Line 50 in 5b6cd5f

ggdist (>= 3.1.0),

[nikosbosse]

My concern would be that it forces people to update to the newest version without a fundamental need. Do you think that is a problem?

[nikosbosse]

My concern would be that it forces people to update to the newest version without a fundamental need. Do you think that is a problem?

[Bisaloo]

No, I think it's perfectly fine. There is a fundamental need to update: to stop using deprecated options.

This is very different from the case where we require a minimum R version. Updating an R package shouldn't be problem for most users while updating R is almost impossible in some very restricted environments.

[nikosbosse]

Makes sense.
I'm not entirely sure we can currently prevent the tests from failing.
When running devtools::test() I'm getting the following warning:

Warning (test-plot_predictions.R:93): plot_predictions() works without median
Using the size aesthietic with geom_ribbon was deprecated in ggplot2 3.4.0.
i Please use the linewidth aesthetic instead.

But I think that is an issue with ggdist and I guess we just have to wait for them to address it.

This is our call to geom_lineribbon, and I don't see any usage of size

ggdist::geom_lineribbon(
        data = intervals,
        aes(
          ymin = lower, ymax = upper,
          # We use the fill_ramp aesthetic for this instead of the default fill
          # because we want to keep fill to be able to use it for other
          # variables
          fill_ramp = factor(range, levels = sort(unique(range), decreasing = TRUE))
        ),
        lwd = 0.4
      )

[codecov]

Codecov Report

Merging #252 (b7f896a) into master (5b6cd5f) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #252   +/-   ##
=======================================
  Coverage   91.32%   91.32%           
=======================================
  Files          22       22           
  Lines        1361     1361           
=======================================
  Hits         1243     1243           
  Misses        118      118           

Impacted Files	Coverage Δ
R/plot.R	97.96% <100.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[nikosbosse]

huh, interesting - they fail on my machine, but not in CI. well then.. Merging!