[R] I want to use Sweave, but only sometimes

[Duncan Murdoch]

On 4/13/2009 2:07 PM, Paul Johnson wrote:
> Does anybody have a workable system to run an Rnw document through
> R-Sweave when necessary, but to just run it through LaTeX if no new R
> calculations are needed? I.e.,  the figures already exist, I do not
> need R to do more work for me, so I send the document straight to
> LaTeX.

It's not exactly what you asked for, but the weaver package goes some 
way towards that, by caching the results of code chunks and not 
recomputing them.

weaver is on Bioconductor, not CRAN, so you can install it using

     source("http://bioconductor.org/biocLite.R")
     biocLite("weaver")

An alternative approach is to save the slow calculations manually, and 
then load() an image in an early chunk, and proceed from there.  But if 
it's your figures that take a long time, that might not help.  I'm not 
sure that weaver caches figures, but I would guess so.

Duncan Murdoch

> 
> I want to leave open the option that I might need to run the document
> through Sweave in the future, so I don't want to just convert the
> document to pure LaTeX format.
> 
> Here's why I think this must be possible. In this list, I saw one of
> you explain this approach to working with figures in Sweave docs:
> 
> Instead of using the automatic figure inclusion approach like this:
> 
> <<testfn2, fig=true>>=
> curve(sin, from = 1, to = 55)
> @
> 
> Do this instead:
> 
> 
> \SweaveOpts{prefix.string=foo/bar}
> 
> <<testfn2, fig=true, include=false>>=
> curve(sin, from = 1, to = 55)
> @
> 
> 
> \begin{figure}
> \caption{My Figure}
> \includegraphics{foo/bar-testfn}
> \end{figure}
> 
> 
> As long as the figures are saved in the directory foo, then LaTeX will
> find them, I think (hope!).  Then if I could tell LaTeX to ignore the
> <<>> ...@ stuff, then it *seems* to me the Rnw file would be processed
> successfully without further hassle.
> 
> I am sorry if this has been asked & answered before, I may not know
> the "magic words" for searching for past solutions.
> 
> pj
>