[R] Avoid Sweave from stopping on errors

[Roger Bivand]

On Wed, 7 Sep 2005, Douglas Bates wrote:

> On 9/7/05, Doran, Harold <HDoran at air.org> wrote:
> > You could use <<eval=FALSE>>= and the code inside the chunk will not
> > be evaluated. I suppose two other options could be to comment out the
> > bad code inside the code chunk or to use verbatim to make it look like
> > a code chunk in your output.
> > 
> > 
> > 
> > -----Original Message-----
> > From: r-help-bounces at stat.math.ethz.ch [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Vincent Goulet
> > Sent: Wednesday, September 07, 2005 11:44 AM
> > To: r-help at stat.math.ethz.ch
> > Subject: [R] Avoid Sweave from stopping on errors
> > 
> > Hi all,
> > 
> > Is there an option in Sweave to avoid it from stopping on a code chunk
> > with an error? (I purposefully want to include code with an error in
> > class notes.)
> > 
> > I suspect the answer is "no" and that I will be pointed to
> > options("error"). That'd be fine, but which error parameter will just
> > "do nothing" in case of an error?
> > 
> > Thanks in advance!
> > 
> 
> Another option is to wrap the call that will produce an error in
> try().  You still get the error report, etc. but execution does not
> stop.

This works very nicely for the original purposes, as the try() can be 
hidden and the command in error echoed:

<<echo=TRUE,eval=TRUE>>=
y <- rnorm(10)
is.na(y) <- 1 
y
@
<<echo=TRUE,eval=FALSE>>=
lm(y ~ 1, na.action=na.fail)
@
<<echo=FALSE,eval=TRUE,results=verbatim>>=
try_out <- try(lm(y ~ 1, na.action=na.fail))
cat(try_out)
@

looks in output as it should, with the error message set after the command 
that provoked it.

Sweave is a remarkable tool.

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no