Behaviour of arranging within groups has changed

[khakieconomics]

Hi,

Having some issues arranging within groups with the dev version. I get that the output may have changed for principled reasons, but I'd just like to know if it's intentional before going off and doing a lot of refactoring. Please see example below.

# Using old version
install.packages("dplyr", lib = "~/Downloads/dplyr_0.4.2.tar.gz")

library(dplyr)

# Create some data
version_1 <- data_frame(a = c("a", "a", "b", "b"), b = c(4,2,3,1)) %>% group_by(a) %>% arrange(b)

# What does it look like? 
version_1


####
# Restart R Session
####

# Get latest dplyr
library(devtools)
install_github("hadley/dplyr")
library(dplyr)

# Create the same data as above
version_2 <- data_frame(a = c("a", "a", "b", "b"), b = c(4,2,3,1)) %>% group_by(a) %>% arrange(b)

# Look at the new version. The ordering has changed so that now the output doesn't respect groups
version_2

# They _are_ the same
all.equal(version_1, version_2)

# But when I convert them to data.frames they're not. Unfortunately we do this quite a bit. 
all.equal(version_1 %>% as.data.frame, version_2%>% as.data.frame)

[hadley]

Yes, this is by design.