| RE: Widgets break CALL_EXT? [message #5728] |
Fri, 02 February 1996 00:00 |
mallozzi
Messages: 60
Registered: August 1994
|
Member |
|
|
> Your problem may be due to IDL's use of unix signals. I ran into a problem
> using CALL_EXTERNAL to perform network i/o. IDL uses signals for a variety
> of tasks. In particular, SIGALRM is used for all timing events. When
> X-windows is used, this timer is set to go off on a regular basis. What
> seems to happen is that if a system call (such as i/o) is done and is
> interrupted by SIGALRM, then the system call will not complete.
-SNIP-
Much thanks to Ken Knighton <knighton@gav.gat.com> for the key to the
solution to this problem. Sure enough, I wrote two little c subroutines
to block SIGALRM and then restore the default state of the handler after
CALL_EXTERNAL was finished, and presto!, everything worked :-)
In my opinion, this behavior is a bug under IDL widgets, and should be
fixed in a future IDL release. Thanks again, Ken
-Bob Mallozzi <mallozzi@ssl.msfc.nasa.gov>
|
|
|
|
| Re: Widgets break CALL_EXT? [message #5739 is a reply to message #5728] |
Thu, 01 February 1996 00:00  |
Ken Knighton
Messages: 44
Registered: May 1995
|
Member |
|
|
mallozzi@ssl.msfc.nasa.gov wrote:
> Hi all,
>
> I believe that there is a problem with CALL_EXTERNAL under some operating
> systems. For a FORTRAN subroutine that requires termial I/O, an error is
> generated if a widget is created before the call to CALL_EXTERNAL. The READ
> statement in the FORTRAN subroutine generates an error. However, if no widgets
> are used, the call succeeds. I did not test with a C routine, but would be
> interested to see if it produces the same error.
>
Your problem may be due to IDL's use of unix signals. I ran into a problem
using CALL_EXTERNAL to perform network i/o. IDL uses signals for a variety of
tasks. In particular, SIGALRM is used for all timing events. When X-windows
is used, this timer is set to go off on a regular basis. What seems to happen
is that if a system call (such as i/o) is done and is interrupted by SIGALRM,
then the system call will not complete. This can cause all sorts of strange
things to happen to your fortran or C code. I got around this problem by
using the sigblock() system call to block the delivery of SIGALRM while my
i/o was completing and then to unblock the signal (HPUX).
The IDL Advanced Development Guide addresses this problem in chapter 10. It
also documents the IDL_TimerBlock() function call (a C function), that will
accomplish the same thing (and probably takes care of other problems) in a
platform independent way.
I hope this helps.
Ken Knighton knighton@gav.gat.com knighton@cts.com
General Atomics
San Diego CA
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
