Date: Tue, 16 Dec 97 12:27:59 EST
From: bhyde@gensym.com (Ben Hyde)
Message-Id: <9712161727.AA13641@gensym1.gensym.com>
To: new-httpd@apache.org
Subject: RE: [PATCH] Serialize the update to pool.sub_* in destroy_pool

: <01BD097F.59A3B740.dda@individual.com>
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@apache.org

"David D'Antonio" <dda@individual.com> wrote:
> If you mean Win32, the CriticalSection stuff works pretty
> well. I believe that NT (not as sure about Win95) will
> still be able to kill off a misbehaving process, even if
> its in a CriticalSection. On the other hand, it *will*
> make it easy for you to deadlock yourself if you aren't
> careful. In a previous life, we used CriticalSections for
> our web server without mishap...

Humm, further investigation indicates that create_mutex
would needs a pool if this is going to work out.  The
CriticalSection needs space.  A cleanup is nobel.

Every thread should have pool, and create_thread should
take a pool to create that pool in.

So now I'm double frightened: NT semantics and the number
of Apache source lines this would effect.

I don't believe I have ever seen a need for topdown clean
up routine.  Debugger like activities often require a
first try exception handler, which is similar, but
different: "pull the fire alarm before doing anything
else" exception handlers.

  - ben h.
---
"Normal people are those whom you don't know very well."