is.null & is.na are missing parentheses

[JohnMount]

Using the RSQLite dplyr adaption of tbl dplyr::mutate I am seeing previous values of the left-hand side of assignments lost in certain situations. I think it has to do with types (int/double/logical) causing some confusion.

Example below (and here):

library('dplyr')
d <- data.frame(x=c(1,2,2),y=c(3,5,NA),z=c(NA,'a','b'),
                rowNum=1:3,
                stringsAsFactors = FALSE)
print(d)

fnam <- tempfile(pattern = "dplyr_doc_narm", tmpdir = tempdir(), fileext = "sqlite3")
my_db <- dplyr::src_sqlite(fnam, create = TRUE)
class(my_db)
dRemote <- copy_to(my_db,d,'d',rowNumberColumn='rowNum',overwrite=TRUE)


# correct calculation
dRemote %>% mutate(nna=0) %>%
  mutate(nna=nna+ifelse(is.na(x),1,0)) %>% 
  mutate(nna=nna+ifelse(is.na(y),1,0)) %>% 
  mutate(nna=nna+ifelse(is.na(z),1,0))  
 
# incorrect calculation (last step seems to always clobber the previous result)
dRemote %>% mutate(nna=0) %>%
  mutate(nna=nna+is.na(x)) %>% 
  mutate(nna=nna+is.na(y)) %>% 
  mutate(nna=nna+is.na(z))

# clean up
rm(list=setdiff(ls(),'fnam'))
if(!is.null(fnam)) {
  file.remove(fnam)
}
gc()

[krlmlr]

Thanks, confirmed. Need to add parens in the SQL rendering of is.null() and is.na(). Would you like to contribute a fix?

[JohnMount]

krlmlr, I'd be thrilled to. Your diagnosis plus dplyr::explain made the problem very clear. My pull request is here: #2307 .

Update: It will be better if somebody else patches this. Sorry about the missteps on my part.