opensubscriber
   Find in this group all groups
 
Unknown more information…

e : expect@listserv.ActiveState.com 6 September 2005 • 7:23AM -0400

Re: [Expect] Expect for windows and DejaGNU
by David Gravereaux

REPLY TO AUTHOR
 
REPLY TO GROUP




Jeff Hobbs wrote:
> Frederic RISS wrote:
>
>>On Thu, 2005-08-25 at 14:58 -0700, Jeff Hobbs wrote:
>>
>>>David Gravereaux wrote:
>>>
>>>>Frederic RISS wrote:
>>>>
>>>>>On Fri, 2005-08-19 at 03:46 -0400, Dave Fisher wrote:
>>>>>
>>>>>>Second, the system call does NOT put the output on the connected
>>>>>>pipes
>>>>>>(stdin/stdout/stderr) of a spawned process because it creates another
>>>>>>process - this is why you are timing out on them.
>>>>
>>>>Oh, subprocess of the spawned app, yes..
>>>>
>>>>I thought that got fixed?  I remember that bug.
>
>
>>Is there any chance this can be corrected ? We badly depend
>>on subprocess and this will make our entire testsuite
>>useless. If it could at least only timeout and not totally
>>crash expect, this would be nice.
>
>
> Dave responded with a possible correction for this bug at:
>
> http://aspn.activestate.com/ASPN/Mail/Message/expect/2795941
>
> I tried this, but did not get the output of the extra child
> processes of my spawned processes.  I suspect there is more
> involved relating to ensuring the child processes are watched
> and controlled in the same way the spawned parent is.
>
>   Jeff Hobbs, The Tcl Guy
>   http://www.ActiveState.com/, a division of Sophos


The answer for it is in there somewhere.  I found the call to
ContinueDebugEvent from a EXIT_PROCESS_DEBUG_EVENT from a child of the
spawned process to cause an exception in the child.  I was using net.exe
for the testing.

I know nothing more..  A possible handle resource thing maybe?

_______________________________________________
Expect mailing list
Expect@list...
http://listserv.ActiveState.com/mailman/listinfo/expect

Bookmark with:

Delicious   Digg   reddit   Facebook   StumbleUpon

Related Messages

opensubscriber is not affiliated with the authors of this message nor responsible for its content.