| Re: Loop Arrays [message #27245 is a reply to message #26977] |
Mon, 15 October 2001 09:35  |
Ken Mankoff
Messages: 158
Registered: February 2000
|
Senior Member |
|
|
On Tue, 9 Oct 2001, Mark Hadfield wrote:
> From: "Ken Mankoff" <mankoff@lasp.colorado.edu>
>> I am interested in creating circular arrays, where subscripts that would
>> be out-of-bounds on a regular array just start indexing on the other side
>> of the array.
> BTW another indexing extension that appeals to me is Python's use of
> negative indices to refer to positions relative to the end of the array,
> eg -1 refers to the rightmost element.
I have a few ideas for implementing this feature. If you are interested in
the option and ability to do this, please let me know your suggestions on
the following.
The algorithm is simple: Evaluate the numbers, variables, and functions
that make up the subscripts w.r.t. the dimensions of the arrays. If any
are negative or larger than their dimension, then re-write code to behave
in new way.
What is needed is a regular expression filter and replacement algorithm.
If any of the following is detected, then replace with:
a[ -3 ] a[ n_elements(a)-3 ]
a[ 3-n ] if( 3-n lt 0 ) then a[ n_elements(a)-3-n ] else a[3-n]
a[ 2, -3 ] a[ 2, (size(a,/dim))[2] -3 ]
a[ 2:-4 ] a[ 2: n_elements(a)-4 ]
a( -3 ) * let normal interpreter handle ()
etc...
I wrote this in IDL ( for about 1/2 the cases) before I realized it has
some pitfalls: retall and .full_reset_session exit the routine, and this
routine tries to emulate $MAIN$, which is bad.
I realized I can fix those problemn with emacs lisp or Perl. With either
one, I can open a pipe between IDL and the user, give them the ability
turn on/off wierd_subscript_method, and then reformat their input so the
IDL interpreter understands it. This could be done in any language, but I
assume that Perl would be best since it is good with regexp and string
manipulations. Does anyone have any other suggestions?
-k.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
