Return-Path: owner-new-httpd Received: by taz.hyperreal.com (8.6.10/8.6.5) id NAA03638; Thu, 8 Jun 1995 13:59:07 -0700 Received: from bauhaus.organic.com by taz.hyperreal.com (8.6.10/8.6.5) with ESMTP id NAA03633; Thu, 8 Jun 1995 13:59:06 -0700 Received: (from cliffs@localhost) by bauhaus.organic.com (8.6.12/8.6.12) id OAA04826 for new-httpd@hyperreal.com; Thu, 8 Jun 1995 14:00:21 -0700 Message-Id: <199506082100.OAA04826@bauhaus.organic.com> From: cliffs@organic.com (Cliff Skolnick) Date: Thu, 8 Jun 1995 14:00:19 PST In-Reply-To: Rob Hartill "Re: More on logging" (Jun 8, 1:46pm) X-Mailer: Mail User's Shell (7.1.1 5/02/90) To: new-httpd@hyperreal.com Subject: Re: More on logging Sender: owner-new-httpd@apache.org Precedence: bulk Reply-To: new-httpd@apache.org Instead of SHMEM, you can use mmap() to simulate shared memory. I'd write the code to go either way, then you cover 100% of modern UNIX variants and many older ones. On Jun 8, 1:46pm, Rob Hartill wrote: } Subject: Re: More on logging } > } > > > If the logging process is the parent, then it should simply be } > > > passing a pointer to a log information structure for the child to } > > > fill in as it handles the request. } > > } > > doesn't that need shared memory ? } > } > I suppose it does. As I mentioned in earlier mail, some of thes } > features probably can't be available to *all* UNIX variants. } > I admit that logging is an important one, but we can always } > leave the standard CLF in place for those. } > } > Are there other ways besides SHMEM to accomplish this? } } Yup, all children drop the info down a socket to the logging process :-) } } -- } Rob Hartill } http://nqcd.lanl.gov/~hartill/ }-- End of excerpt from Rob Hartill