Dev versions of ggmosaic and ggplot2 clash - why?

[shippy]

See explanation, reprex, and session info at haleyjeppson/ggmosaic#11. ggmosaic does something bad, breaking ggplot2 internals. The only clue is the fact that "as.tibble.list" is getting printed via cat, which only appears in this file, and that unloading tibble is the only way to get ggplot functional again.

While this is primarily a ggmosaic/tibble issue, the fact that ggplot2 is breakable in this way might be useful to know and guard against.

[shippy]

Okay, I think what's happening is that ggmosaic is monkey-patching tibble::as_tibble.list S3 method, while also loading tibble via Depends rather than Imports. I didn't realize you could monkey-patch methods in R like that; I have no idea if there's anything to be done to protect from that, other than not loading ggmosaic.

While this particular issue should be closed, there's probably an unreported issue about incorrect ggplot2 / tibble handling of productplot data frames (or something). If I get a chance to explore this next week, I'll file that issue if I can pinpoint the problem.