Problem in extracting a subset of a ddf

[talegari]

Hi,

I am trying to extract a subset of a ddf(local disk based).

library("datadr")
write.table(mtcars, "mtcars.csv", sep = ",", row.names = FALSE)
lc <- localDiskConn(file.path(getwd(), "test"), autoYes = TRUE)
drRead.table("mtcars.csv"
             , header = TRUE
             , sep = ","
             , output = lc
             , rowsPerBlock = 7)
mtcars_ddf <- ddf(lc, update = TRUE)
drSubset(mtcars_ddf, gear == 2)

The last line throws an error:

* Testing 'subset' on a subset
Error in names(tmp) <- c("key", "value") : 
  'names' attribute [2] must be the same length as the vector [1]

Same error if the last line is replaced by: drSubset(mtcars_ddf, c(rep(TRUE, 10), rep(FALSE, 22)))
Is this the intended usage of drSubset or am I missing something?

[hafen]

Sorry for the delay. This is interesting. What version of datadr and R do you have? I don't get an error, but a row of all NA (which isn't right either - it should be an empty data frame).

The other test case you provide, where you provide a vector of logicals, doesn't work in this setting because a ddf doesn't keep track of the row ordering of the original data frame.

The intended usage of drSubset() is a bit limiting because it's intended to be used when you want to grab an in-memory data frame subset of a ddf, which could potentially be very large.

If you want to filter out records but still have a ddf, you can simply add a transformation function to a ddf that does the filtering (using whatever mechanism you'd like - subset, dplyr, etc.).

[talegari]

I am now getting a row of NA's now, an empty dataframe is preferable though. This issue may be closed, if required.