In the past few years we have been working on making it possible to use packages in both R and Spotfire S+ (previously officially called S-PLUS). One problem was that certain lines in the DESCRIPTION file of a package needed to be specific to R or Spotfire S+. E.g., if your package used the xyplot() function then the package would depend upon lattice in R and trellis in Spotfire S+, or a package might require R 2.8.0 or S+ 8.1 to work properly. Another example might be that R's build system requires that the Maintainer: line contains an e-mail address but in Spotfire S+ a URL is acceptable.
Our solution to this is that if a DESCRIPTION file has entries tagged 'Xyx:' and 'XyzSplus:' then S+'s read.dcf() function will ignore the 'Xyz:' entry and output the 'XyzSplus:' entry as the 'Xyz:' entry. E.g., the Depends: example above would be handled by putting the following 2 lines in DESCRIPTION Depends: R (>= 2.8.0), lattice DependsSplus: Splus (>= 8.1), trellis When R's read.dcf() reads that DESCRIPTION file the Depends component of its output would be "R (>= 2.8.0), lattice" and when S+'s read.dcf() reads it the Depends component would be "Splus (>= 8.1), trellis". R's read.dcf() would also return the component "DependsSplus", but that should not hurt anything as long as the Depends component were also present (otherwise partial matching could cause x$Depends to give you x$DependsSplus). The Maintainer: issue would be handled by Maintainer: Joe the Package Writer <[EMAIL PROTECTED]> MaintainerSplus: Acme Support <http://acmewidgets.com> S+'s read.dcf() has an argument, convertSplusLines=TRUE, that one can use to avoid this translation (useful mainly when copying DESCRIPTION files by using write.dcf(read.dcf(oldfile),newfile)). This should not affect how things work in R (unless a package writer had a tag ending with 'Splus') but package writers may find it handy or may wonder why we add such lines when we help port their packages to Spotfire S+. This feature is new to Spotfire S+ 8.1. Bill Dunlap TIBCO Spotfire Inc wdunlap tibco.com ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
