Errors with . and .data pronouns in do()

[hongooi73]

This is with a fresh build of dplyr as of 20 minutes ago. It looks like multiple references to . or .data is confusing the evaluator.

# this works
> do(mtcars, m=lm(mpg ~ ., data=.))
# A tibble: 1 x 1
         m
    <list>
1 <S3: lm>


# these don't work
> do(mtcars, m=lm(mpg ~ ., data=.), w=lm(wt ~ ., data=.))
Error in is.data.frame(data) : object '.' not found

> do(mtcars, m=lm(mpg ~ ., data=.data), w=lm(wt ~ ., data=.data))
Error in is.data.frame(data) : object '.data' not found

[krlmlr]

I suspect this is due to the forceful cleaning of the data mask in eval_tidy_(), which is called by do() in a loop. (I changed the implementation of do() to rebuild the overscope for each item inside map(), that fixed the bug.) eval_tidy_() is now deprecated. What's the best way to proceed? Is #3318 related?

[lionel-]

Yes with previous rlang it should have been overscope_eval_next() rather than eval_tidy_() as the former does not clean the mask.

With the new simplified API you just create an overscope/data mask and supply it to eval_tidy(). Note that there is an issue with rlang 0.2.0 that will be fixed in 0.2.1, the mask is still cleaned up at each call. So until the next version is out I would use the soft-deprecated overscope_eval_next().

[krlmlr]

Thanks, Lionel, your suggestion seems to work. Simpler reprex:

dplyr::do(mtcars, a = nrow(.), b = nrow(.))
#> Error in nrow(.): object '.' not found

Created on 2018-03-19 by the reprex package (v0.2.0).

[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/