Anova mermod fix

[bbolker]

This addresses #1114

[bbolker]

Ugh, needs more work. I made some unwarranted assumptions.

[simonpcouch]

Thanks for getting this started!

Ugh, needs more work. I made some unwarranted assumptions.

I felt this.🫠 Working with this output can be really tricky—I wish we had some more attributes or classes to work with.

Let me know if you'd appreciate me spending some time here! Glad to review whenever feels good to you.

[bbolker]

I could solve this problem on the lme4 end by making anova.merMod() return an object with class equal to c("anova.mer", "anova", "data.frame"), and writing a separate method for it to be included in broom.mixed ... would that be better? (The only downside is that it would presumably copy a lot of code from the tidy.anova() method ...)

[simonpcouch]

That sounds like a good approach! re: the copying code downside, much of the tidy.anova source is now the Special catch for car::linearHypothesis conditional (which I presume you wouldn't need?) and I could export the renamers object as anova_tidier_renamers so that that vector would automatically be synced up. Another downside, though, is that this would require users to load broom.mixed in a place where they used to be able to get by with only the broom package loaded, and the reason for that isn't visible in the sense that output from the anova() generic could require either package depending on the method dispatched to.

Another option, which is less expressive but is more backward-compatible and likely more maintainable, suggested in review:

[github-actions]

This pull request has been automatically locked. If you believe the issue addressed here persists, please file a new PR (with a reprex: https://reprex.tidyverse.org) and link to this one.