| Re: strange behaviour of bytscl by large arrays [message #80096 is a reply to message #79963] |
Sat, 05 May 2012 04:39   |
Lajos Foldy
Messages: 176
Registered: December 2011
|
Senior Member |
|
|
On Friday, May 4, 2012 5:46:39 PM UTC+2, Chris Torrence wrote:
> Hi all,
>
> I've added a fix for this in IDL Next (not 8.2). If you ask for values larger than 16777215, then it switches to using a 64-bit integer for the counting, and then casts those values to floats. You will still get duplicate numbers and discontinuities (impossible to fix that), but at least it won't get stuck on a single number. I did the same thing for double, in case you have any desire to allocate 9007199254740992 elements (that's 8 petabytes).
>
> I also added a START keyword to all of the *INDGEN routines and MAKE_ARRAY, that allows you to specify a starting index. I needed this for testing purposes so I didn't have to keep creating huge arrays, and it seemed like a generally useful feature. It's much faster to specify START than to add an offset to the indgen after you've created it.
>
> Cheers,
> Chris
> ExelisVIS
Thanks.
May I suggest SEED or ROOT instead of START? START will break 'make_array(5, /st)' :-)
regards,
Lajos
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
