Strip .data from default labels

[hadley]

@thomasp85 this ensures that when you use the .data pronoun with ggplot2, it doesn't adversely affect the labels.

@paleolimbot — this should make your talk slightly simpler 😄

[thomasp85]

Shouldn't strip_dots() get the plotenv passed into it with this new setup?

[hadley]

I don't think the plotenv should matter here, since all the environment information is stored in the quosures inside the aesthetics. The function signature is mildly confusing because the default env should never actually be used since it's always overridden by the quosure environment; it's only need to get recursion off the ground.

[thomasp85]

If you feel certain about this then fine... but leave a comment about it because we usually use plotenv instead of globalenv

[hadley]

If you're concerned about it, I can just leave it without a default — if it fails (which based on my understanding of the code, shouldn't happen), it'll be caught by the tryCatch() so you'll just get a slightly less optimal label.

[thomasp85]

I have no problem with it, but it is just slightly unclear that evaluation happens in the correct environment without a comment. So for future generations of me, I'd like a small comment

[thomasp85]

If you remove the default, the argument is basically unused all the way through the recursion. Why have the argument at all then? I feel I'm missing something about this code...

[thomasp85]

Ah... now I see. Disregard my last comment