Yes, I can confirm ggplot2 has changed default binning. This difference is why it is listed as a breaking change.

Factors and strings are treated at discrete

It'll be properly supported instead of accidentally in the next version.

Paranormal distribution

Gotcha, nah I don't know of any package implementing these specifically. I'd be surprised if nobody would have implemented something along these lines though.

The go-to function for me would be `scales::col_mix()`. Is that what we're looking for?
scales.r-lib.org/reference/co...

I think this is in the right direction. Maybe solar output?

marquee.r-lib.org/reference/gu...

No no, I don't love boxplots. Boxplots have blindsplots too. I like showing individual data points (as long as they're tractable). You can throw in whatever summaries too, but eyeballing the data is important :)

Yeah using complete themes provided by extension packages is a great way of using themes. I mentioned the ggthemes package as well for this!

I empathise some people might be affected by this change! Hopefully the rest of the features and bugfixes make up for it!

Dynamite plots hurt the soul 💔

The reason is that ggplot2 can't keep on accumulating a lot of new features, or it risks becoming so big it'll be unmaintainable. It has a great system that allows others to extend its functionality, which is also how the legendry package works.

`ggsave(bg)` is doing the exact same thing. The difference you might see is that themes now adhere to the `ink`/`paper` settings. This also populates the `plot.background` setting, which wasn't always the case before.

It sets the device's background colour. If you have a non-transparent `plot.background` setting, it will show on top of the device's background colour.