| "foreach" loops in IDL [message #64806] |
Fri, 16 January 2009 10:30  |
alaniwiusenet
Messages: 2
Registered: January 2009
|
Junior Member |
|
|
In case anyone's interested, I've produced a little hack that lets you
have "foreach" loops in IDL, i.e. loop over the contents of an array.
It would be nice if the syntax were to let you have something like:
for myval in myarray do begin
print, myval
endfor
Unfortunately it doesn't, unless I've missed something somewhere. And
it's just a little fiddly to have to loop over an array index every
time, e.g.:
for i=0, n_elements(myarray) - 1 do begin
myval = myarray[i]
print, myval
endfor
The hack lets you use the following syntax:
foreach, 'myval', myarray
@do
print, myval
@done
and nested loops are supported.
If anyone is interested in having this, grab this a (tiny) download,
that contains a few short files to put somewhere in your IDL_PATH:
http://home.badc.rl.ac.uk/iwi/idl-foreach.tar.gz
Regards,
Alan
P.S. Please note that I do not read this mailbox. Do a web search for
"Alan Iwi" if you want my email address.
|
|
|
|
|
|
| Re: "foreach" loops in IDL [message #64867 is a reply to message #64806] |
Tue, 20 January 2009 14:57  |
Vince Hradil
Messages: 574
Registered: December 1999
|
Senior Member |
|
|
On Jan 20, 3:11 pm, JD Smith <jdtsmith.nos...@yahoo.com> wrote:
> On Jan 16, 3:43 pm, Vince Hradil <vincehra...@gmail.com> wrote:
>
>> I would have to agree with Paolo here. The for-loop syntax is simple
>> and clear. The @do @done really obfuscates the code.
>
> That's because it's an unsupported add-on. If IDL included foreach at
> the language level, which would be clearer:
>
> for i=0,n_elements(x)-1 do begin
> elem=x[i]
> print, my_function(elem)
> end
>
> or
>
> foreach elem in x
> print, my_function(elem)
> end
>
> When treating a vector as a list, requiring an extra loop variable is
> pure syntactic overhead. Not to mention that the loop variable could
> overflow, could get changed in the body of the loop or, most commonly,
> risks nested sub-loops accidentally re-using the same loop variable.
> None of these happens with a foreach construct.
>
> JD
Oh, I agree - a foreach loop would be great to have in IDL. My only
point was that add-on is confusing to read, and I would prefer to
write more portable code that uses as much built in functionality as
possible.
|
|
|
|
| Re: "foreach" loops in IDL [message #64868 is a reply to message #64806] |
Tue, 20 January 2009 14:54 |
rtk
Messages: 22
Registered: September 2008
|
Junior Member |
|
|
On Jan 20, 2:11 pm, JD Smith <jdtsmith.nos...@yahoo.com> wrote:
> foreach elem in x
> print, my_function(elem)
> end
>
> When treating a vector as a list, requiring an extra loop variable is
> pure syntactic overhead. Not to mention that the loop variable could
> overflow, could get changed in the body of the loop or, most commonly,
> risks nested sub-loops accidentally re-using the same loop variable.
> None of these happens with a foreach construct.
Exactly! Of course, at this point, we are talking Python:
for elem in x:
print my_function(elem)
Lists are a natural extension/complement to IDL's arrays. Then, once
you have lists, you will naturally want higher-order functions to
operate on those lists. Lists would add great power to IDL but
perhaps if people are not familiar with lists they will not appreciate
what lists have to offer. To prove my point, look at what happened
with Python. It supports lists out of the box but folks added
powerful array processing via the numpy package. Well, IDL is just
the opposite. It has powerful array processing already, now it just
needs lists.
The package I referenced above has a list DLM that is quite useful on
its own. It supports all IDL data types except structures. A truly
integrated version would naturally add helpful syntax and support all
data types.
Ron
(Disclaimer: while I do work for ITT VIS, these opinions are my own
and should not be thought of as those of ITT VIS...)
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
