[R] installing my own package - problems with INDEX

Martin Maechler maechler at stat.math.ethz.ch
Tue Jun 15 15:13:16 CEST 2004


>>>>> "Pamela" == Pamela Hall <phall at alum.mit.edu>
>>>>>     on Tue, 15 Jun 2004 08:41:27 -0400 writes:

 <...............>

    Pamela> I must be doing something else wrong.  Its not the braces.

    Pamela> This is what I have done that may not be right?

    Pamela> 1.  I put all of the Rd files into 1 file with a \eof separating them.  This is how it appears in already built packages, however, I haven't seen any reference to that in the R manual.   But if I don't do that, R CMD check completely fails.

That's definitely not the recommended way.
But it might not be the problem either


    Pamela> 2.  What about the functions themselves?  I have tried them as separate files or concatenated into one file.  The functions always appear to have been sourced and are available in R, but the help pages fail either way.

as you're guessing yourself, this should be a completely
orthogonal matter.

    Pamela> 3.  \keyword.  Very few of my functions really fit into any of the keyword.db options.  I note that there also appears to be keywords for packages that can be added (eg spatial).  Can I make my own keyword and add it to the keyword.db?

No, don't.

This is becoming a FAQ..  (we had this topic recently with Rmetrics),

The \keyword{}s are not ``key words'' in the usual sense, but
rather  __category__ entries. Categories of applied statistics /
data analysis.  To choose one, you might find it easier to read
KEYWORDS rather than KEYWORDS.db

For ``real'' key words, you can nowadays use \concept{ .. }
the content of which is also searched by help.search() [when the
package is installed].

Again, please tell us what you are doing and we may help you to
choose appropriate keywords (or --much less probably-- you
convince us that some new keywords are needed).

Regards,
Martin Maechler




More information about the R-help mailing list