On Mon, Oct 18, 2004 at 07:06:31PM -0400, Jesse Pavel wrote:
>
>
> I'm trying to structure my Quixote SCGI program so that each child
> process started by the SCGI server acquires a few resources when it
> begins, and explicitly releases them when the parent process
> terminates. I'm not sure where is the proper place to put the code
> that releases resources, considering that I normally terminate the
> parent with a SIGINT.
>
> When I do that, what happens to the child processes, which are, I
> presume, blocking while waiting for the passed file descriptor from
> the server? Is there a "clean" method of releasing resources, based
> on catching a Python exception somewhere, or perhaps I need to trap
> a signal in each child?

Have you implemented and are you using a sub-class of
quixote.publish.Publisher?  That's what I do.  My Publisher
subclass usually contains:

1. A start_request method, which is called just before request
   processing.  It shoves the resources my request processing needs
   into the request object.

2. A __del__ method, which is called when the process and the
   Publisher are destroyed.  It cleans up my resources.

Of course, you will want to put a debugging message in the __del__
method to make sure that it is being called.  As I recall, there is
a bit of weirdness about the __del__ method.  A print statement
placed directly in the __del__ method works.  But, a call to
another function containing a print statement does not.  At least,
that's what I recall.

Dave


--
Dave Kuhlman
http://www.rexx.com/~dkuhlman