Skip to content

Checking package vignettes: WARNING

5 messages · Pfaff, Bernhard Dr., Friedrich Leisch, Arne Henningsen

#
Hi,

I want to have a "vignette" in one of my R packages. Hence, I added an Sweave 
file into the /inst/doc subdirectory of this package. Unfortunately, 'R CMD 
check' gives a warning:

==========================================
[...]
* checking tests ... OK
* checking package vignettes in 'inst/doc' ... WARNING
--- f?hre texi2dvi auf Vignetten aus
* creating sampleSelection-manual.tex ... OK
* checking sampleSelection-manual.tex using pdflatex ... OK

WARNING: There was 1 warning, see
  /home/suapm095/Documents/Econometrics/R/micEcon/micEcon/sampleSelection.Rcheck/00check.log
for details
==========================================

Since the line below "WARNING" is in German, I ran 'LC_ALL="C" R CMD check' 
to get an English message (which I could use in a Google search). However, the 
warning does NOT occur if I run 'R CMD check' in the 'C' locale. I tried a 
lot of different things and I figured out that 'f?hre texi2dvi auf Vignetten 
aus' is the translation of 'running texi2dvi on vignettes', but I did not 
find the reason for the warning message that occurs if 'R CMD check' is run 
in the German locale. If I run Sweave and texi2dvi by hand with both locales, 
the only possibly relevant difference I noticed is that Sweave uses straight 
single quotation marks in the 'C' locale but (italicized) primes in the 
German locale:

'C' locale:
   Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

German locale:
   Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1 

System Information:
R 2.7.1
i486-pc-linux-gnu
Kubuntu 8.04 

Any hints are welcome!
Thanks,
Arne
#
Hello Arne,

I do not know if this matters, but have you used 'results = verbatim' in your R code chunk? See page 13 in: http://www.statistik.lmu.de/~leisch/Sweave/Sweave-manual.pdf

Best,
Bernhard
*****************************************************************
Confidentiality Note: The information contained in this ...{{dropped:10}}
#
Hi Bernhard!
On Wednesday 16 July 2008 14:16:05, you wrote:
Thank you for this hint. I have tried this option, but it seems to have no 
effect on the quotation marks. Hence, I still do not know whether the 
quotation marks or something else causes the warning.

Arne

  
    
#
> Hi Bernhard!
> On Wednesday 16 July 2008 14:16:05, you wrote:
>> I do not know if this matters, but have you used 'results = verbatim' in
  >> your R code chunk? See page 13 in:
  >> http://www.statistik.lmu.de/~leisch/Sweave/Sweave-manual.pdf

  > Thank you for this hint. I have tried this option, but it seems to have no 
  > effect on the quotation marks. Hence, I still do not know whether the 
  > quotation marks or something else causes the warning.


verbatim is the default, so adding changes nothing if you had no
results=xxx before.  

  >> >German locale:
  >> >
  >> >'C' locale:
  >> >   Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  >> >
  >> >German locale:
  >> >   Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  >> >
  >> >System Information:
  >> >R 2.7.1
  >> >i486-pc-linux-gnu
  >> >Kubuntu 8.04
  >> >


In whiche locale are you running? On Debian Linux with de_AT.UTF-8
everything is fine for me.

Of course you also need something like

\usepackage[utf8]{inputenc}

to handle the special characters, but you need that anyway because
every textual output of R may contain Umlaute ...

Hth,
Fritz
#
On Wednesday 16 July 2008 18:28:07, Friedrich Leisch wrote:
Thanks for this hint! (I did not have any 'results=xxx' before. Hence, it is 
clear why adding 'results = verbatim' had no effect.)
de_DE.UTF-8
Yes, I have this command in the preamble of the Rnw file, but I get the 
waning message anyway :-(

I will uploaded the sampleSelection package (version 0.6-1) to 
   ftp://cran.r-project.org/incoming
If the CRAN servers run in the 'C' locale, all tests should pass without a 
warning and the updated package with the vignette will be available on CRAN 
soon. In this case, everybody can download the package and try to figure out 
the reason for the warning message. Any hints are welcome! 

BTW: It would be great if "R CMD check" could return a more informative 
warning message.

Thanks,
Arne