dot in dcast formula - anything goes

[Henrik-P]

Disclaimer: this is basically me doing a mistake, but because it may raise a more general question -- if an argument value not described in the help text should result in an error -- I thought I just as well could share it.

My intention was to achieve the functionality of . (". represents no variable") in the dcast formula. But, I had forgot the correct syntax and just throwed in a 1 instead (I might have thought of the 'no variable' / intercept only in lm...).

The 1 in the formula gave the desired (dot-)result, but when I saw the . in the variable name I realized my mistake. Should it be possible to use argument values not described in the help text?

# some data
d <- data.table(x = rep(LETTERS[1:2], each = 4),
                y = rep(letters[1:2], each = 2),
                z = 1:8)

# correct dot code, which I had forgot
dcast(d, x + y ~ ., value.var = "z", fun.aggregate = sum)

# my erroneous code, with a 1 instead of a dot gives the same result
dcast(d, x + y ~ 1, value.var = "z", fun.aggregate = sum)
 #   x y  .
# 1: A a  3
# 2: A b  7
# 3: B a 11
# 4: B b 15

# just to try what may translate to a dot - anything..;)
dcast(d, x + y ~ "anything", value.var = "z", fun.aggregate = sum)
 #   x y  .
# 1: A a  3
# 2: A b  7
# 3: B a 11
# 4: B b 15