[Rd] R CMD build/check

[Kurt Hornik]

>>>>> Paul Gilbert writes:

> Let me start out by saying that after using "R CMD build" for a few days
> it appears to be a truely wonderful facility, even if I didn't really
> want to know about all the problems with my packages right now.

>> Please try to eliminate all problems reported by the R package builder
>> before submitting a package to CRAN.

> Now here are a few caveates. Please let me know if  I'm not aware of
> something that has already been fixed.

> 1/ I'm using a file 00General_Description.Rd to provide a general
> description of the package. This gets reported as an Rd file without
> alias, but even though the package builder reports that as a problem I
> didn't intend to eliminate it.

The rule is as follows (from the R Extension Writers guide):

  `\alias{TOPIC}'
     The `\alias' entries specify all "topics" the file documents.
     This information (together with the file name) is collected into
     index data bases for lookup by the on-line (plain text and HTML)
     help systems.

     There may be several `\alias' entries.  Quite often it is
     convenient to document several R objects in one file.  For example,
     file `Normal.Rd' documents the density, distribution function,
     quantile function and generation of random variates for the normal
     distribution, and hence starts with

          \name{Normal}
          \alias{dnorm}
          \alias{pnorm}
          \alias{qnorm}
          \alias{rnorm}

     Note that the name of the file is not necessarily a topic
     documented.

Hence, an Rd file without an alias is pointless because it does not
register any topic it provides information for.

> 2/ Specific methods get reported as undocumented functions unless the
> generic documentation has an alias line for the specific method. This
> is not a problem when everything is in one package, however, for many
> of my specific methods the generic method is in another of my packages
> or in the base. (This can generate a lot of warning, which obscure
> more important warnings.) I'm putting a "stub" description in to
> eliminate all the warnings, but I'm not sure if this is the ideal way
> to do this.

Not sure if I understand this correctly.  If you have a specific method
say foo.dse then you'd have

  \alias{foo.dse}

etc.  So why would this be undocumented?

> 3/ Data sets cannot use constructors in the package. They need to be
> structures which can be fully recognized by the base system. (.rda
> files may be a better alternative in this case?)

Yes.  I will add something on this to the R-exts.

> 4/ R CMD build does not work for bundles, but it can be used to test
> sub-packages and then the bundle rolled up afterwards directly with
> tar and gzip. (This is not difficult and not a problem, but people
> should be aware of it.)

I should add something on this too.  Perhaps, Paul, once you release dse
as a package bundle I can try to make R CMD build work for bundles, too.

-k
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._