| Re: IDL 8.0 compile_opt changes [message #69108] |
Mon, 21 December 2009 09:23  |
chris_torrence@NOSPAM
Messages: 528
Registered: March 2007
|
Senior Member |
|
|
Just a quick note about the negative indices. They are *only* allowed
in ranges or as a scalar index. They are *not* allowed in arrays of
indices.
For example:
a[0:-1] In IDL80, returns all the elements
a[-1] In IDL80, returns the last element
a[[-1,0,1,2]] will either truncate the -1 to 0, or will throw an
error if the "strictarrsubs" compile option is set. This behavior will
remain the same in IDL80. We will not change it.
So, for example, David's example:
z= where(bad ne 0)
if z[0] ne -1 then begin
lots of stuff
endif
This will still work *fine* in IDL80.
Now, in IDL80 the following might cause a problem:
z= where(bad ne 0)
do something with bad[z] (without checking for -1)
In this case, if "where" returns a scalar -1, then bad[z] will indeed
return the last element. However, it would be bad to use the where
result without checking for a -1...
So, in summary, in IDL 8.0:
1. Negative indices are only used for ranges or a scalar index
2. Negative indices within an index array will still behave the same
(either clip or throw an error)
3. The behavior of Where remains unchanged (and you still need to
check for a -1 result, as always)
Returning to the compile_opt discussion...
-Chris
ITTVIS
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
