[Rd] R CMD check failing to warn when it should

Martin Maechler maechler at stat.math.ethz.ch
Fri Aug 5 14:18:48 CEST 2005

>>>>> "DeepS" == Deepayan Sarkar <deepayan.sarkar at gmail.com>
>>>>>     on Wed, 3 Aug 2005 13:52:32 -0500 writes:

    DeepS> Hi, I recently made changes to lattice code which
    DeepS> needed changes in many man pages as well. Before I
    DeepS> made the appropriate changes, R CMD check was
    DeepS> flagging most of the problems correctly, except for
    DeepS> the man page for tmd. I have created a toy package
    DeepS> that shows this, available at

    DeepS> http://www.stat.wisc.edu/~deepayan/R/tmdprob_0.12-2.tar.gz

    DeepS> This passes R CMD check on R 2.1.0 and r-devel from
    DeepS> August 1, but it shouldn't because the code and
    DeepS> documentation are inconsistent.

It's because you use \synopsis{}.  
This basically breaks all 'codoc' checking and is the reason
we (mainly Kurt, but I completely agree with him) have been
thinking about deprecating its use -- possibly using a
substitute for the few cases that might need something like it.

Kurt and I (at least) would very strongly advocate not to use
\synopsis{}, and hence writing functions and methods in a way
that can be well documented with exact \usage{....}.


More information about the R-devel mailing list