I have 3 questions about creating a package for R, and would appreciate any guidance. 1. Demonstrating the packaged code requires some support code files. It would seem natural to put these in the demo directory, but "Writing R extensions" says files in demo should be suitable for runing from demo(). The support files are not intended to be run from the top level. How should I handle them? A further complication is that the demo won't actually work without some prior setup outside of R, setup that can't be automated (the user must pick a computational cluster). I am planning to handle that by documenting it. 2. Are there some base datasets I can assume present for use in my demonstration? 3. Is this the right list for these questions? Thanks. -- Ross Boylan <ross at biostat.ucsf.edu> wk: (415) 502-4031 University of California, San Francisco fax: (415) 476-9856 Dept of Epidemiology and Biostatistics hm: (415) 550-1062 530 Parnassus Avenue (Library) rm 115-4 San Francisco, CA 94143-0840
On 6 May 2003, Ross Boylan wrote:> I have 3 questions about creating a package for R, and would appreciate > any guidance. > > 1. Demonstrating the packaged code requires some support code files. It > would seem natural to put these in the demo directory, but "Writing R > extensions" says files in demo should be suitable for runing from > demo(). The support files are not intended to be run from the top > level. How should I handle them?Perhaps put them in inst/> A further complication is that the demo won't actually work without some > prior setup outside of R, setup that can't be automated (the user must > pick a computational cluster). I am planning to handle that by > documenting it. > > 2. Are there some base datasets I can assume present for use in my > demonstration?The ones listed by data() in a plain installation of R. Nearly all installations will also have the data() sets from the recommended packages, and you can use these if you check (eg with require()) that the package is actually present.> 3. Is this the right list for these questions?It's not an egregiously wrong list. r-devel at r-project.org might be better. -thomas