Re: [Rd] Issue when calling deparse(substitute(x)) from C with "anonymous" R vectors ?

From: Peter Dalgaard <P.Dalgaard_at_biostat.ku.dk>
Date: Thu, 19 Nov 2009 09:37:05 +0100

Laurent Gautier wrote:
> Dear list,
>
>
> When calling R from C, what appears like a spurious error can be
> triggered during the execution of chisq.test(x, y).
>
> This is happening when the following conditions are met:
>
> - x and y are "anonymous" C-level R vectors (they do not have a symbol),
> but they are protected from garbage collection
>
> - x and y are "not too small" (it was experienced as soon as they are
> longer than 17 elements).
>
>
> The error is
>
> Error in names(dimnames(x)) <- DNAME :
> 'names' attribute [4] must be the same length as the vector [2]
>
> and can be traced to the use of deparse(substitute(x)) and
> deparse(substitute(y)) in the R code.
>
> What seems to be happening is that the deparse(substitute(x)) call
> gives a character vector of length > 1 as soon as x is "not so small".
>
> To demonstrate this, I am using rpy2 (as the problem was found when
> using it
> http://sourceforge.net/mailarchive/forum.php?thread_name=4B043CA1.9050500%40salilab.org&forum_name=rpy-list
>
> ), but it will likely be present in other bridges to R as well.
>
>
> #using R-2.10 and rpy2-2.1.dev

>>>> import rpy2.robjects as robjects

>
>>>> f = robjects.r('''function(x) return(deparse(substitute(x)))''')

>
>>>> tuple(f(robjects.FloatVector(range(17))))

> ('c(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)',)
> # length 1
>>>> tuple(f(robjects.FloatVector(range(18))))

> ('c(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17', ')')
> # length 2 !!
>
>
> Does it seem to others like an issue present in other bridges as well ?
>

It can be generated from R itself:

> x <- sample(0:1,2000,replace=T)
> y <- sample(0:1,2000,replace=T)
> eval(substitute(chisq.test(x,y),list(x=x,y=y)))
Error in names(dimnames(x)) <- DNAME :
  'names' attribute [252] must be the same length as the vector [2]

Looks like nlines=1 is needed in the deparse().

-- 
   O__  ---- Peter Dalgaard             Ă˜ster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark      Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard_at_biostat.ku.dk)              FAX: (+45) 35327907

______________________________________________
R-devel_at_r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Received on Thu 19 Nov 2009 - 08:39:24 GMT

This archive was generated by hypermail 2.2.0 : Thu 19 Nov 2009 - 09:40:30 GMT