R devel - Nov 2009 - Issue when calling deparse(substitute(x)) from C with "anonymous" R vectors ?

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 ?



L.

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             ?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

This is easy to reproduce in R:

chisq.test(c(1000, 1001, 1002, 1003, 1004, 1005, 1006, 1007, 1008, 1009, 1010,
1011),
c(1000, 1001, 1002, 1003, 1004, 1005, 1006, 1007, 1008, 1009, 1010, 1011))

The simple answer is: don't do that.

It is unclear what is a reasonable label to give in such a case: maybe 
simply 'x' and 'y'?

On Thu, 19 Nov 2009, 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 ?
>
>
>
> L.
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

Can you put together a minimal, self contained example, not requiring an 
external system?  This has the symptoms of a bug in the external code 
(e.g. a missing PROTECT), but it is possible it's a bug in R.

Duncan Murdoch

On 19/11/2009 2:40 AM, 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 ?
> 
> 
> 
> L.
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel