R devel - Nov 1999 - Dimnamenames (PR#257)

I've just started digging into Andreas's old report:
> a <- array(0,c(2,2,2));
> dimnames(a)<-list(hansi=1:2, pepi=c("a","b"),
karli=3:4)
>  dimnames(a[,,1]) # losing component names
[[1]]
[1] "1" "2"

[[2]]
[1] "a" "b"

Now, this would be fairly easily fixed (I think) which would also make
it compatible with S3. But what should one do about a[,1,1]?  This
would generally be a vector with a .Names attribute, onto which there
is no way of attaching a name. The obvious idea would be to force it
to be a 1d array instead, i.e.

structure(c(0,0), .Dim = 2, .Dimnames =
structure(list(hansi=c("1","2"))))

rather than

structure(c(0,0), .Names = c("1","2"))

Does this sound safe? Generally, such 1-d arrays behave like vectors,
and noone has reported trouble with the fact that the result of table
can be a 1d array. From some viewpoints it seems odd to have different
behaviour depending on whether something is named or not, but it
*would* be nice to have the label retained on rows extracted from a
table...

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Hi Peter,

On Wed, 3 Nov 1999 p.dalgaard@biostat.ku.dk wrote:
> Now, this would be fairly easily fixed (I think) which would also make
> it compatible with S3. But what should one do about a[,1,1]?  This
> would generally be a vector with a .Names attribute, onto which there
> is no way of attaching a name. The obvious idea would be to force it
> to be a 1d array instead, i.e.
> 
> structure(c(0,0), .Dim = 2, .Dimnames =
structure(list(hansi=c("1","2"))))
> 
> rather than
> 
> structure(c(0,0), .Names = c("1","2"))
> 
> Does this sound safe?
I'm not sure.  Aren't you duplicating the functionality provided by the
drop flag?  After all, if the names were important then presumably the
subscription would have been a[,1,1, drop=FALSE] anyhow.  I like the
concept of a vector in this situation and suspect that in my lazy code I
have stuff that checks for a dim attribute whan checking arguments to
functions, which would be a problem.  Perhaps it would be better to pass
the name as a different attribute, eg

structure(c(0, 0), .Names = c("1","2"), .dnm =
"hansi")

But it would be great to have the names fixed!

Cheers, Jonathan.

Jonathan Rougier                       Science Laboratories
Department of Mathematical Sciences    South Road
University of Durham                   Durham DH1 3LE

"[B]egin upon the precept ... that the things we see are to be 
 weighed in the scale with what we know"  (Meredith, 1879, The Egoist)



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._