FR for support commandArgs

[jangorecki]

While argv is somehow clear way it requires the R script to be adjusted to be called from littler. littler could hook the commandArgs function and inject own argv it would allow running scripts originally written for Rscript - more portable.

[eddelbuettel]

littler predates Rscript. I have little interest in this.

Plus, if you use a proper tool like docopt for R then it is handled for you.

[jangorecki]

Thanks for info. Instead of new dependency the following would if(!exists("argv")) argv = commandArgs(TRUE) allows portability of R scripts.
Leaving for reference:

if(!exists("argv")) argv = commandArgs(TRUE) else argv = as.character(argv)
cat(str(argv))
q("no")

as.character is added to match case of no input argument where argv would be NULL while commandArgs(T) would be character(0).

[eddelbuettel]

Thanks. I stuck that onto the wiki.

[eddelbuettel]

I added some comments on the wiki. The more I think about it, the less sense it makes.

r always has argv but not commandArgs(TRUE):

edd@max:~$ r -e'print(argv); print(commandArgs(TRUE))' a b c
[1] "a" "b" "c"
character(0)
edd@max:~$ 

Did you mean to assign the other way?

[jangorecki]

I mean a R code which will works the same for both r and Rscript.

[eddelbuettel]

Should the setting of argv then be added to Rscript instead?

[jangorecki]

Yes, but changing Rscript behaviour seems to be much harder than adjusting each R script to work for both r and Rscript.
argv seems to be way to go as it is widely adopted outside of R already. Should I fill FR for it somewhere? in bugzilla??

[eddelbuettel]

Not sure. You may want to discuss it on r-devel first as you will need a sponsor for the change in any event.