file.path about efficientr HOT 9 CLOSED

In my experience file.path is only useful in specific cases where there's a conflict between Windows and Unix-alike OS paths. In this case I suggest we don't use it - will remove I get the green light...

from efficientr.

I'm happy with either.

What about mentioning file path in an rmdnote

from efficientr.

Sounds good - will do

from efficientr.

FWIW, I think the main thing worth mentioning is that if you're calling an external program, depending on how that program behaves you might need Windows \ slashes vs. / slashes (and so normalizePath() is your friend).

At the R level, using paths with / (or from file.path()) is generally the right thing to do; it's just munging those paths for e.g. calls to system() that can be tricky.

from efficientr.

@kevinushey Thanks, very helpful.

from efficientr.

One thing that IMHO is not covered in enough detail by Hadley's 'R Packages' is where to put raw files (pick from /inst/exdata, /vignettes, /testthat/tests), but also how to deal with finding these files. For example, when building a vignette, the folder /vignette is temporarily uses as a working directory, so a path like /inst/exdata/my_file.txt will fail there.

As an ugly workaround I end up writing functions like this:

find_path <- function(filename) {
  if (file.exists(filename)) {
    return(filename)
  }

  prefixes <- c(
    "E:/Git/Cer2016/inst/extdata/",                                    # nolint
    "/home/richel/GitHubs/Cer2016/inst/extdata/",                      # nolint
    "/home/p230198/GitHubs/Cer2016/inst/extdata/",                     # nolint
    "/home/travis/build/richelbilderbeek/Cer2016/inst/extdata/",       # nolint
    "C:/Users/Aline/Cer2016/inst/extdata/"                             # nolint
  )
  for (prefix in prefixes) {
    full_path <- paste(prefix, filename, sep = "")
    if (file.exists(full_path)) {
      return(full_path)
    }
  }
  stop(
    "find_path: ",
    "cannot find '", filename, "'"
  )
}

I hope you guys come up with some clear guidelines on this. If you haven't already written something up one this :-)

from efficientr.

Thanks. I've managed to somehow avoid this problem, so I don't much insight. Perhaps @Robinlovelace does.

I'm not sure if this fits naturally into the book, but we'll have a think.

from efficientr.

Just been looking into this. Seems inst/extdata is the right place according to the first 3 results of an internet search, the R Packages book (which says "If you want to store raw data, put it in inst/extdata") and this question

Here is a typical example, from the datapack package:

csvfile <- system.file("extdata/sample-data.csv", package="datapack")

@kevinushey I'll add your insight and fix this asap.

Many thanks for the input.

from efficientr.

Awesome! Thanks for the research!

from efficientr.