Sexpr values from external chunk not found

[fonnesbeck]

I have a large external R script, which I am trying to incorporate via knitr's read_chunk function. However, for some reason none of the variables referred to in the surrounding text are able to be found. I have confirmed that the R script runs cleanly on its own, so it is not a case of the script failing before variables are defined. I had a look to see if others have had similar problems, but could not find anything, so I assume I am doing something wrong.

I am reading in the file using read_chunk in a setup section:

<<set-options, echo=FALSE, cache=FALSE>>=
read_chunk("study.R")
@

And below that, I use pieces of the script that I have annotated as follows:

<<km, echo=FALSE, results='hide'>>=
@

So, to give one example, there is an object d2 that is defined in the script within the km section, and in the text below, I refer to it as follows:

There are at present \Sexpr{sum(d2$recur)} total events (recurrences or deaths), and \Sexpr{sum(d2$death)}
 deaths out of \Sexpr{nrow(d2)} patients randomized.

However, I end up with errors as follows:

Error in eval(expr, envir, enclos) : object 'd2' not found
Error in eval(expr, envir, enclos) : object 'd2' not found
Error in nrow(d2) : object 'd2' not found

In fact, there is not a single instance of a Sexpr statement being properly processed. Any idea what I am doing wrong?

[yihui]

so you have a line like ## @knitr km in study.R? and what is your sessionInfo()?

[mnel]

Do you use plyr or reshape2

If so, it could be related to #207,

you may need to update to the most recent github version of reshape2

see #18.

[fonnesbeck]

I do not use plyr or reshape2 for this project.

I do indeed have knitr markup in my code. For example:

## @knitr "km"

## Estimate KM curves
pdf("Fig/kmGI.pdf", width=10, height=10)
par(mfrow=c(2,2),las=1)

etc.

Here is my sessionInfo after running the script:

> sessionInfo()
R version 2.14.1 (2011-12-22)
Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] graphics  grDevices utils     datasets  stats     methods   base     

other attached packages:
[1] knitr_0.3.4   ggplot2_0.9.0

loaded via a namespace (and not attached):
 [1] codetools_0.2-8    colorspace_1.1-1   dichromat_1.2-4    digest_0.5.1       evaluate_0.4.1    
 [6] formatR_0.3-4      grid_2.14.1        highlight_0.3.1    MASS_7.3-16        memoise_0.1       
[11] munsell_0.3        parser_0.0-14      plyr_1.7.1         proto_0.3-9.2      RColorBrewer_1.0-5
[16] Rcpp_0.9.10        reshape2_1.2.1     scales_0.2.0       stringr_0.6        tools_2.14.1 

[mnel]

Try

## @knitr km

I tried on a simple example, and removing the quotation marks worked

[fonnesbeck]

This did not do the trick, unfortunately. Is there a problem with referencing variables from a previous chunk? For example, I have:

<<base, echo=FALSE, results='hide'>>=
@

\section{Pre-planned Analysis (All data)}

\subsection{Kaplan Meier Estimates}

<<km, echo=FALSE, results='hide'>>=
@
%\newpage
There are at present \Sexpr{sum(d2$recur)} total events (recurrences or deaths), and \Sexpr{sum(d2$death)}
 deaths out of \Sexpr{nrow(d2)} patients randomized. 

...

where the object d2 is defined in base rather than in km. That should not matter, right? Yet, I get:

  |>>>>>>>>>>                                                       |  16%
   inline R code fragments

Error in eval(expr, envir, enclos) : object 'd2' not found
Error in eval(expr, envir, enclos) : object 'd2' not found
Error in nrow(d2) : object 'd2' not found
Error in eval(expr, envir, enclos) : object 'ncut' not found 

[mnel]

Hmmmm. Are d2 etc in the global environment when you finish knitting the document?

If you have cache=T then have you deleted the cache files to force everything to be reevaluated.

[fonnesbeck]

They are not when I finish knitting the document, but they are if I just run the R script on its own.

I am not cacheing anything, so that should not be the issue.

[yihui]

Can you remove echo=FALSE, results='hide' for the chunk km and check if this chunk is evaluated at all? I mean check the tex output.

Although I do not think the version matters for this problem, the current knitr version on CRAN is 0.5 now, so you may update it before moving on.

[fonnesbeck]

I think I discovered the problem: d2 was defined in a chunk of code previous to "km" that was not given a knitr label. I made a label at the beginning of the file, and now d2 etc. are seen. Sorry for the confusion. I did not realize that all portions of the code needed to have a @knitr section label.

[yihui]

You are welcome. Glad you sorted it out finally!

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue by following the issue guide (https://yihui.org/issue/), and link to this old issue if necessary.