[Bioc-devel] \donttest and the "80% of man pages documenting exported objects must have runnable examples" rule

Richard Cotton richierocks at gmail.com
Tue May 17 14:23:36 CEST 2016

On 17 May 2016 at 13:43, Martin Maechler <maechler at stat.math.ethz.ch> wrote:
> I also wonder *why*  you use \donttest{} so extensively.
> It's clearly better than  \dontrun{}  (which really distresses
> me, if used more than occasionally).

I totally agree with you that \dontrun is overused and annoying for users.

My main use of \donttest is to speed up R CMD check.  Further up the
thread I mentioned that this package is related to file import, and
that many of the examples take a second or two to run, since they
import a text file included in the package.  Since many of the
examples begin with the same code importing the sample file,
repeatedly testing it isn't that beneficial given the time it takes to
run them.

I've been mostly persuaded by Hervé that it would be better to use
--no-examples when calling R CMD check for this purpose.  That way I
can turn testing the examples on and off in one place.


Learning R

More information about the Bioc-devel mailing list