In WinVector/replyr: Patches to Use 'dplyr' on Remote Data Sources

Factor with R data.frame.

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = " # "
)
options(width =100)

Some issues with summarize_each and factors.

library('dplyr')
R.Version()$version.string
packageVersion('dplyr')
d1 <- data.frame(y=c('a','b'),stringsAsFactors = FALSE)
d1 %>% dplyr::summarise_each(dplyr::funs(lexmin = min,lexmax = max))
d2 <- data.frame(y=c('a','b'),stringsAsFactors = TRUE)
d2 %>% dplyr::summarise_each(dplyr::funs(lexmin = min,lexmax = max))

Submitted as dplyr issue 2269. Closed as "expected behavior" as this is what min(factor(letters)) does. That is a correct determination, but be aware many dplyr backends do support comparison, min, and max on characters types.

my_db <- dplyr::src_sqlite("replyr_sqliteEx.sqlite3", create = TRUE)
dplyr::copy_to(dest=my_db,df=d1,name='d1',overwrite=TRUE) %>% 
  dplyr::summarise_each(dplyr::funs(lexmin = min,lexmax = max))
dplyr::copy_to(dest=my_db,df=d2,name='d2',overwrite=TRUE) %>% 
  dplyr::summarise_each(dplyr::funs(lexmin = min,lexmax = max))