[R] do.call() and aperm()

[Tony Plate]

Thanks, I appreciate knowing that.

abind() can currently take a fractional value for along, and behaves as per 
your description of 'catenation' in APL.

Does APL supply any hints as to what sort of value to give 'along' to tell 
abind() to perform 'lamination'?

-- Tony Plate

At Tuesday 01:22 PM 10/21/2003 -0400, Gabor Grothendieck wrote:


>I suggest following APL as that is a well thought out system.
>In APL terms there are two operations here called:
>
>- catenation. In abind, this occurs when along = 1,2,...,length(dim)
>- lamination.  In abind, this occurs when along = length(dim) + 1
>
>however, the latter is really only one case of lamination in
>which the added dimension comes at the end.  To do it in full
>generality would require that one can add the new dimension
>at any spot including before the first, between the first and
>the second, ..., after the last.
>
>In APL notation, if along has a fractional part then the new
>dimension is placed between floor(along) and ceiling(along).
>Thus along=1.1 would put the new dimension between the first
>and second.  The actual value of the fractional part is not material.
>
>---
>From: Tony Plate <tplate at blackmesacapital.com>
>
>I've also been thinking about how to specify that 'along' should be
>length(dim)+1. At the moment one can specify any number from 0 up to
>length(dim)+1, but as you point out you have to spell out length(dim)+1 as
>the value for the along argument. It would possible to make abind()
>automatically calculate along=length(dim)+1 when given along=NA, or
>along=-1, or along="+1". Any preferences?
>
>-- Tony Plate
>
>
>
>_______________________________________________
>No banners. No pop-ups. No kidding.
>Introducing My Way - http://www.myway.com