Some tests in downstream packages that use the ggplot object fail

[paleolimbot]

This is a revdep issue (#3303) related to #3310 in that the theme is the thing that caused the plot object to change from the previous version, but different in that the downstream problem is how to properly test ggplot output.

MTLR does not define a complete theme, but has a .rds file containing a previous theme object that does not match the current ggplot theme objects (since they have all changed).

ggstatsplot tests its custom theme_pie() using numeric indices ( https://github.com/IndrajeetPatil/ggstatsplot/blob/master/tests/testthat/test_theme_ggstatsplot.R#L65 ). The custom theme is created using theme_bw() + theme(...) and is probably fine, but the index of the object they're trying to test has changed since an element was inserted.

CSTools has a test for the length of the theme object ( https://github.com/cran/CSTools/blob/master/tests/testthat/test-PlotForecastPDF.R#L13 ).

[clauswilke]

All of these are bad practices and our course of action should be to contact the maintainers. Could you reach out to them?

[paleolimbot]

Can I suggest that folks use the vdiffr package? Or will that be hard because of #3304?

[paleolimbot]

Sorry, just figured out that vdiffr doesn't run on CRAN. I'll have these issues filed shortly.

[paleolimbot]

• MTLR Forthcoming ggplot2 release and MTLR haiderstats/MTLR#1
• ggstatsplot Forthcoming ggplot2 release and ggstatsplot IndrajeetPatil/ggstatsplot#213
• CSTools isn't on GitHub, I emailed the maintainer

[clauswilke]

Thanks, I think we can close this issue.

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/