Skip to content
Prev 15281 / 63421 Next

How graphical an interface should the default be?

Very nice additions you have made! Thanks.
Minor comment 1: Isn't ask=(TRUE|FALSE) and graphics=(TRUE|FALSE) more in
line with the other updates? True, one more argument, but you can imagine
update.packages(ask=TRUE) with a "fall-back" to a text-based selection menu.
Until implemented, one could default to graphics=ask.

Minor comment 2: Naming conventions, not discussed very often, but I would
suggest chooseCranMirror() instead of chooseCRANmirror(). (I comment on this
in under "3.1 General Naming Conventions" -> "Abbreviations and acronyms
should not be uppercase when used as name" in my RCC draft at
http://www.maths.lth.se/help/R/RCC/.)
I would say that same/identical behaviour on as many platforms as possible
should be favored. That will make it easier to write general
instructions/help and have subsection with OS-specific features.
My feeling is that a global option is what you want. Not only in the above
cases, it would be nice for any package developer to be able to query such
an option. I also believe there are different users; some prefer GUIs some
prefer command line, or even strongly dislike GUIs).

If writing scripts for, say, automated installation of 450+ CRAN packages on
various systems such Linux, Windows and OS X, it would be very useful to
have a predictive behaviour of the above so that you can make it work the
same on all systems. For example, I used to redefine readline() to
automatically answer questions asked by install.packages() when installing
many packages at the same time; this would be much harder if a GUI pops up. 

Maybe there should be a new options to start R with, e.g. --nogui. 

Best

Henrik Bengtsson