R devel - Jan 2003 - empty array indexing with dimnames (PR#2507)

On Thu, 30 Jan 2003 maechler@stat.math.ethz.ch wrote:
> This is a bug that will not often trigger, and only happens for arrays
> (but not matrices) ending up with 0 dimensions:
The bug is a missing else branch of an if condition at line 431 of 
subset.c.

I don't think the buglets are wrong.  You don't drop length 0
dimensions.
If you did, the object would be of the wrong length.  You only drop length 
1 dimensions.
> 
> The following code also shows two 'buglets' (the first even for
> matrix indexing) where "drop = TRUE/FALSE" is not correctly
obeyed.
> One can argue that since all these arrays have length zero, it
> shouldn't matter what dimensions they get.
> [But this shouldn't distract from the real bug giving errors
>  where it shouldn't!]
> 
> 
> ##-- this kind of indexing works with matrices (with dimnames):
> m <- cbind(x = 1, y = 1:3)
> str(m[1,  0])# drops dimension
> str(m[1:2,0])# doesn't drop  <<< ? buglet
> str(m[1,  0, drop=FALSE])
> str(m[1:2,0, drop=FALSE])
> 
> 
> ## these apply (not too relevant here)
> stopifnot(identical(m[1,         0],
>                     m[1, integer(0)]))
> stopifnot(identical(m[1,         0, drop=FALSE],
>                     m[1, integer(0), drop=FALSE]))
> 
> ###--- but it fails with arrays ---
> 
> dn <- list(LETTERS[1:2], letters[1:3],
paste("t",1:4,sep=""))
> A. <- array(1:24, dim = 2:4) # no dimnames
> Ad <- A.; dimnames(Ad) <- dn # with dimnames
> 
> ## works `ok', if no dimnames --- not always obeying "drop="
--> 2nd buglet!
> for(O in list(0, integer(0))) {
>   str(A.[1, O, 2 ]) # int(0)
>   str(A.[O,-1, O ])# int[0, 1:2, 0]
>   str(A.[O, O,2:3])# int[0, 0, 1:2]
> 
>   ## with drop = FALSE:
>   str(A.[1, O, 2 , drop=FALSE])# int[1, 0, 1]
>   str(A.[O,-1, O , drop=FALSE])# int[0, 1:2, 0]
>   str(A.[O, O,2:3, drop=FALSE])# int[0, 0, 1:2]
>   cat("---\n")
> }
> 
> ##-- identical with `Ad' instead of `A.', i.e., *with* dimnames:
> ##--> All 6 statements fail :
> 
> O <- 0##__for(O in list(0, integer(0))) {
> 
>   try(Ad[1, O, 2 ]) # int(0)
>   try(Ad[O,-1, O ])# int[0, 1:2, 0]
>   try(Ad[O, O,2:3])# int[0, 0, 1:2]
> 
>   ## with drop = FALSE:
>   try(Ad[1, O, 2 , drop=FALSE])# int[1, 0, 1]
>   try(Ad[O,-1, O , drop=FALSE])# int[0, 1:2, 0]
>   try(Ad[O, O,2:3, drop=FALSE])# int[0, 0, 1:2]
>   cat("---\n")
> ##__}
> 
> ______________________________________________
> R-devel@stat.math.ethz.ch mailing list
> http://www.stat.math.ethz.ch/mailman/listinfo/r-devel
> 
-- 
Brian D. Ripley,                  ripley@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

>>>>> "BDR" == Brian D Ripley
<ripley@stats.ox.ac.uk>
>>>>>     on Thu, 30 Jan 2003 23:49:22 +0100 (MET) writes:
    BDR> On Thu, 30 Jan 2003 maechler@stat.math.ethz.ch wrote:
    >> This is a bug that will not often trigger, and only
    >> happens for arrays (but not matrices) ending up with 0
    >> dimensions:

    BDR> The bug is a missing else branch of an if condition at
    BDR> line 431 of subset.c.

and you have already fixed it for 1.7.0

    BDR> I don't think the buglets are wrong.  You don't drop
    BDR> length 0 dimensions.  If you did, the object would be
    BDR> of the wrong length.  You only drop length 1 dimensions.

Yes, of course, you are right!

     <....>

    BDR> -- Brian D. Ripley, ripley@stats.ox.ac.uk Professor of
    BDR> Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
    BDR> University of Oxford, Tel: +44 1865 272861 (self) 1
    BDR> South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG,
    BDR> UK Fax: +44 1865 272595

Martin Maechler <maechler@stat.math.ethz.ch>
http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><