[R] Sweave problem with recursive call
g.wrobel@dkfz-heidelberg.de
g.wrobel at dkfz-heidelberg.de
Fri Jul 12 11:33:53 CEST 2002
Hi,
while trying to play around with Sweave I found out that I was unable to call
Sweave in a code chunk that is being parsed by Sweave.
Code chunk that is being parsed:
<<>>=
for (i in 1:length(chips[,1])) {
createAnalysis(as.character(chips[i,1]),i)
}
system(paste("./concatenateAnalysis -",i,sep=""))
@
The function createAnalysis that is called:
createAnalysis <- function(curChip, counter) {
write(curChip,"direct.chip")
Sweave("analyseChip.Rnw")
system(paste("mv analyseChip.tex aC_",counter,".tex",sep=""))
}
And the code chunk from analyseChip.Rnw:
<<>>=
gpr <- gpr.load.file(curChip,gpr.load.std)
@
\begin{description}
\item[Filename] \Sexpr{curChip}
\item[Date] \Sexpr{gpr$DateTime}
\end{description}
That leads to the follwing error:
> require(tools)
Loading required package: tools
[1] TRUE
> Sweave("analyse.Rnw")
Writing to file analyse.tex
Processing code chunks ...
1 : echo term verbatim
2 : echo term verbatim
3 : echo term verbatim
Read 1 items
Read 6144 records
Read 1 records
Error: can not set length of non-vector
Error in driver$runcode(drobj, chunk, chunkopts) :
Error: can not set length of non-vector
Basically Sweave creates a text connection to an object called "output" in the
main environment while evaluating a code chunk. When calling Sweave in a code
chunk this "output" is referenced with several textconnections. I attached a
diff file with a suggestion how to circumvent the problem. It can be applied
to the Sweave.R file in the tools-1.5.0.tar.gz archive.
Greetings
Gunnar Wrobel
-------------- next part --------------
286c286,292
< tmpcon <- textConnection("output", "w")
---
> if (is.finite(max(as.integer(sub("output","",apropos("output"))),na.rm=TRUE))) {
> oc <- max(as.integer(sub("output","",apropos("output"))),na.rm=TRUE)+1
> } else {
> oc <- 0
> }
> tmpcon <- textConnection(paste("output",oc,sep=""), "w")
> output <- get(paste("output",oc,sep=""))
More information about the R-help
mailing list