| Re: IDL Bug: NCOMPLEMENT keyword to where [message #28030] |
Fri, 16 November 2001 15:58 |
hradilv.nospam
Messages: 19
Registered: November 2001
|
Junior Member |
|
|
sparc sunos unix 5.4 also
On Fri, 16 Nov 2001 15:23:16 -0500, JD Smith
<jdsmith@astro.cornell.edu> wrote:
>
> I've found an ugly bug with the NCOMPLEMENT keyword to where(), first
> introduced in IDL v5.4.
>
> IDL> wh=where(1,NCOMPLEMENT=ncnt)
>
> causes a segfault, in v5.4 and v5.5, under Linux at least. It only
> occurs if NCOMPLEMENT is used, but the corresponding COMPLEMENT is not.
> A temporary workaround is always to specify COMPLEMENT, even if you
> don't need it.
>
> JD
|
|
|
|
| Re: IDL Bug: NCOMPLEMENT keyword to where [message #28032 is a reply to message #28030] |
Fri, 16 November 2001 13:47  |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
On the Mac, the following output is produced:
IDL> wh=where([1,2] eq 1) ; Works fine
IDL> wh=where([1,2] eq 1,NCOMPLEMENT=ncnt)
% Internal error: Bad variable type encountered in no_name_var().
% Execution halted at: $MAIN$
I guess, this confirms the bug.
Pavel
JD Smith wrote:
>
> I've found an ugly bug with the NCOMPLEMENT keyword to where(), first
> introduced in IDL v5.4.
>
> IDL> wh=where(1,NCOMPLEMENT=ncnt)
>
> causes a segfault, in v5.4 and v5.5, under Linux at least. It only
> occurs if NCOMPLEMENT is used, but the corresponding COMPLEMENT is not.
> A temporary workaround is always to specify COMPLEMENT, even if you
> don't need it.
>
> JD
|
|
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
