can we combine two functions? about comparing-sentiment HOT 4 CLOSED

Hm, I'm a bit confused by this, also from the manual, which seems to suggest you cannot combine the above two functions into one function:

Targets with the "file" format are called dynamic files, and these targets are unusual because their return values do not get saved in the _targets/objects/ folder. Rather, they expect you to return a character vector files and directories, which gets stored in the _targets/meta spreadsheet, and then targets watches the data at those paths for changes.

from comparing-sentiment.

But does section 4.3 of the manual suggest that we could create an imported function that both loads (or saves) the file, but still returns a file path.

from comparing-sentiment.

Good point! You can save targets objects as external files and return their path, but then you need to re-read in that path in order to reference these objects inside the pipeline at later target steps. In short: If you want to keep referencing data frames without an additional read-in step, you need to make these two steps. At some points it makes sense to save targets objects externally but I think its not the best idea to do it at all steps. Lastly, I think that tar_target(ss_scale_file, read.table(here("data-sentiment", "sentistrength_scale.txt"), sep="\t", header = T, quote=""), format="file"), would not work since a data object would be returned by the type expects a path/string. Will touch on that in the meeting later today!

from comparing-sentiment.

from comparing-sentiment.