This is a feature suggestion. My apologies if it's already been raised. I recently noticed that dplyr's behaviour has changed:
*dplyr 0.4.0* df <- data_frame(x = c(1, 1, 1, 2, 2), y = 1:5) result <- df %>% dplyr::distinct(x) %>% sum *# result == 8* *dplyr 0.5.0* df <- data_frame(x = c(1, 1, 1, 2, 2), y = 1:5) result <- df %>% dplyr::distinct(x) %>% sum *# result == 3*
That's fine. It's nice to see package maintainers fixing bugs and improving functionality. What's not fine is that for most users, myself included, this change was essentially silent. I would suggest amending the `library` and `require` functions as follows, e.g.:
library(dplyr, version='0.4.0', versioning='semantic') library(dplyr, version='0.4.0', versioning='exact')
If the version check indicates the package has been updated in a way which doesn't maintain backwards compatibility, a warning message would be generated or, possibly, the code would halt. This would provide R users a way of specifying the conditions under which their scripts produced the results they obtain. Best regards, Richard