[Rd] Building R package: make pdf & _masked_by_GlobalEnv

[Dan Rabosky]

>>
>>I am assembling an R package (under Windows XP, R v.2.3) and have a very 
>>basic question.  Running R CMD build does not generate a pdf.  R CMD 
>>check generates a .dvi, but I cannot figure out how to automatically 
>>create the pdf.  I thought from reading the "writing r extensions' manual 
>>that R CMD build was supposed to generate the pdf.  I am having no 
>>success using various permutations of the Rd2dvi and texi2dvi, and I'm a 
>>bit frustrated.  I have miktex 2.4, which should (I thought) have the 
>>necessary pdftex tools to do this.  Maybe it does; I definitely have 
>>various directories named 'dvipdfm', 'pdftex', etc on my computer, but I 
>>have no idea whether there might be a problem.
>
>You need to give a bit more detail.  What is the .pdf being created 
>from?  What do things look like when you run a build?  Does install work?

I guess I'm not sure what the .pdf is being created from - I don't know how 
this works.  I assumed that my set of .Rd files from the man dir would 
automatically be converted into .pdf, much like 'check' reads them into a 
dvi document.   When I run a build, I get an archive with all of my .R 
source files, the DESCRIPTION, the .Rd files, my datafiles, and any 
documentation I've placed in inst/doc.  When I install this from CMD, the 
html help & chtml files compile, examples are copied from the /man 
directory to their own .Rd files, and everything - to me - appears 
OK.  'Install' from R CMD works with no problem - it warns me that my 
'examples' directory is empty, but it proceeds to create the examples from 
my .Rd files anyway.

I only get a .dvi manual when running 'check'.


>>When I install the package, I receive the message that most of my 
>>functions are "masked _by_ .GlobalEnv".  I have no idea what this means; 
>>I'm sure that this is amply documented, but I confess that I have a 
>>rather difficult time with much of the documentation on writing packages.
>
>This means that you have variables in your workspace with the same names 
>as ones in your package.  You probably want to clean out your workspace, 
>because most uses will find those versions before the ones in the package 
>-- and that really messes up testing, etc.

thanks for the clarification!