| obj_destroy on IDL_IDLBridge hangs sometimes [message #72073] |
Thu, 12 August 2010 13:43  |
Alfred de Wijn
Messages: 2
Registered: July 2008
|
Junior Member |
|
|
Hi,
I've been playing around with IDL_IDLBridge objects lately.
Everything seems to be working fine until I try to destroy the
bridges. Here's some code that demonstrates the problem:
ncpus = 4 ; or any other number you like
bridges = objarr(ncpus)
for i=0,999 do begin
print, 'building bridges...'
for cpu=0,ncpus-1 do bridges[cpu] = obj_new('IDL_IDLBridge')
print, 'burning bridges...'
for cpu=0,ncpus-1 do obj_destroy, bridges[cpu]
print, 'loop '+string(i)+' done'
endfor
end
On my Linux systems (both 32 and 64-bit), this will eventually hang
while calling obj_destroy. It appears that loop number 27 is
typically the one where things go wrong, but I've seen it fail earlier
too. When ncpus is larger, e.g., 16, failure is pretty much
guaranteed in the first loop. Does anyone know what's going on? Is
this a known issue?
I'm not particularly familiar with IDL object internals, so it's
likely I'm doing something stupid. Any help appreciated.
Thanks,
Alfred
|
|
|
|
| Re: obj_destroy on IDL_IDLBridge hangs sometimes [message #72240 is a reply to message #72073] |
Fri, 20 August 2010 06:24  |
d.rowenhorst@gmail.co
Messages: 10
Registered: August 2008
|
Junior Member |
|
|
On Aug 20, 9:20 am, "d.rowenho...@gmail.com" <d.rowenho...@gmail.com>
wrote:
> On Aug 12, 4:43 pm, Alfred de Wijn <agdew...@gmail.com> wrote:
>
>
>
>
>
>> Hi,
>
>> I've been playing around with IDL_IDLBridge objects lately.
>> Everything seems to be working fine until I try to destroy the
>> bridges. Here's some code that demonstrates the problem:
>
>> ncpus = 4 ; or any other number you like
>> bridges = objarr(ncpus)
>> for i=0,999 do begin
>> print, 'building bridges...'
>> for cpu=0,ncpus-1 do bridges[cpu] = obj_new('IDL_IDLBridge')
>> print, 'burning bridges...'
>> for cpu=0,ncpus-1 do obj_destroy, bridges[cpu]
>> print, 'loop '+string(i)+' done'
>> endfor
>> end
>
>> On my Linux systems (both 32 and 64-bit), this will eventually hang
>> while calling obj_destroy. It appears that loop number 27 is
>> typically the one where things go wrong, but I've seen it fail earlier
>> too. When ncpus is larger, e.g., 16, failure is pretty much
>> guaranteed in the first loop. Does anyone know what's going on? Is
>> this a known issue?
>
>> I'm not particularly familiar with IDL object internals, so it's
>> likely I'm doing something stupid. Any help appreciated.
>
>> Thanks,
>> Alfred
>
> I can't help, but will verify that I am having the same issue on Mac
> 10.6.4.
Couple of other things I have noticed:
The Output='' keyword option in creating a IDL_IDLBridge also appears
not to work in IDLDE, but works in the command line IDL. Also, I
think that OBJ_DESTROY appears to work in the command line IDL (I have
not thoroughly tested this).
|
|
|
|
| Re: obj_destroy on IDL_IDLBridge hangs sometimes [message #72241 is a reply to message #72073] |
Fri, 20 August 2010 06:20 |
d.rowenhorst@gmail.co
Messages: 10
Registered: August 2008
|
Junior Member |
|
|
On Aug 12, 4:43 pm, Alfred de Wijn <agdew...@gmail.com> wrote:
> Hi,
>
> I've been playing around with IDL_IDLBridge objects lately.
> Everything seems to be working fine until I try to destroy the
> bridges. Here's some code that demonstrates the problem:
>
> ncpus = 4 ; or any other number you like
> bridges = objarr(ncpus)
> for i=0,999 do begin
> print, 'building bridges...'
> for cpu=0,ncpus-1 do bridges[cpu] = obj_new('IDL_IDLBridge')
> print, 'burning bridges...'
> for cpu=0,ncpus-1 do obj_destroy, bridges[cpu]
> print, 'loop '+string(i)+' done'
> endfor
> end
>
> On my Linux systems (both 32 and 64-bit), this will eventually hang
> while calling obj_destroy. It appears that loop number 27 is
> typically the one where things go wrong, but I've seen it fail earlier
> too. When ncpus is larger, e.g., 16, failure is pretty much
> guaranteed in the first loop. Does anyone know what's going on? Is
> this a known issue?
>
> I'm not particularly familiar with IDL object internals, so it's
> likely I'm doing something stupid. Any help appreciated.
>
> Thanks,
> Alfred
I can't help, but will verify that I am having the same issue on Mac
10.6.4.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
