Received: by taz.hyperreal.com (8.6.12/8.6.5) id UAA19713; Thu, 1 Feb 1996 20:48:30 -0800 Received: from fiction.isdn.uiuc.edu by taz.hyperreal.com (8.6.12/8.6.5) with ESMTP id UAA19698; Thu, 1 Feb 1996 20:48:20 -0800 Received: (from blong@localhost) by fiction.isdn.uiuc.edu (8.6.12/8.6.12) id WAA07122 for new-httpd@hyperreal.com; Thu, 1 Feb 1996 22:49:53 -0600 Message-Id: <199602020449.WAA07122@fiction.isdn.uiuc.edu> Subject: Re: port to the BeBox? To: new-httpd@hyperreal.com Date: Thu, 1 Feb 1996 22:49:52 -0600 (CST) In-Reply-To: <199602020341.WAA17572@volterra.ai.mit.edu> from "Robert S. Thau" at Feb 1, 96 10:41:56 pm From: Brandon Long X-Uri: X-Citement: Pontiac X-Disclaimer: I said what? MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1560 Sender: owner-new-httpd@apache.org Precedence: bulk Reply-To: new-httpd@hyperreal.com Last time, Robert S. Thau uttered the following other thing: > > > 2) sockets are not inherited across fork() calls. I see this as being a > bigger problem. > > Yup... it's a problem. There are three common styles of Unix servers > that I'm aware of: > [Good description of current servers deleted] You did miss the threaded servers and async servers, but nothing big is doing that yet. (Paul Phillips Boa, and the Spyglass server, though I'm not sure if its threaded or not (its available for SunOS, which doesn't have threads, but it could be using pthreads for all I know)). > All of these involve transferring permissions on sockets between processes, > either by traditional Unix-style inheritance (as in the first and second > approaches), or more recherche methods. > > Still, there's got to be *some* way to write servers on a BeBox (one hopes!) > --- perhaps they're all threaded, or perhaps there's some incantation which > can cause sockets to be transferred among processes even if it doesn't > necessarily happen automatically. I think the thread model is the prefered way of doing things. If memory serves, they even have a "threaded" windowing system where every window has its own thread. Supposed to be amazingly responsive because of the 2 processors. Brandon -- Brandon Long "I think, therefore, I am confused." -- RAW HTTPd/SDG/NCSA "Of all the things I've lost, I miss my mind the most." ECE/UIUC blong@uiuc.edu http://www.uiuc.edu/ph/www/blong N9WUC Don't worry, these aren't even my views.