From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Message-ID: <43ED0A9E.50404@easysw.com> Date: Fri, 10 Feb 2006 16:50:22 -0500 From: Michael Sweet MIME-Version: 1.0 Subject: Re: [Printing-architecture] Request for comment to PCM draft References: <8A33346A3922BD4EB48A18128BAB16AB474F55@SWA6204.apo.epson.net> In-Reply-To: <8A33346A3922BD4EB48A18128BAB16AB474F55@SWA6204.apo.epson.net> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit List-Id: Printing architecture under linux List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Ide Kentaro Cc: printing-architecture@freestandards.org Ide Kentaro wrote: > Hi, folks, > > I posted link of PCM draft document to OpenPrinting web site. > > > I would like you to comment these documents with two viewpoints. > > 1. Architecture > If you predict that a problem occurs to your system with this architecture, > point out the problem, please. 1. Architecture Diagram It isn't clear how this will work with CUPS; the architecture diagram references an "Open Printing System", but the diagram does *not* match up with how CUPS interfaces with and discovers devices. Basically, I'd focus on the PCM design/architecture itself, and then show possible interfaces with other apps like CUPS, SANE, and so forth. All of the PCM stuff should be in the middle, possible apps/users at the top, and devices at the bottom. 2. Device Discovery The spec and API only focus on accessing a particular device. There is no way to discover or enumerate devices, which is most definitely going to be needed. 3. Asynchronous I/O, Non-Blocking IO, and/or Timeouts The read and write APIs need to support non-blocking I/O or at least basic timeouts so that an application can communicate without getting hung/deadlocked. It would also be useful to get a file descriptor or some other mechanism so that you can poll() or select() for ok-to-read/write conditions. Without this, you are forced to use threading which is problematic. > 2. License > My questions are "Which license is better for us?" and > "As for we should we select the license that was unified > with all the Open Printing documents? ". > > We chose BSD like license to PCM specification, today. > But some other documents chose another license as FDL. > One of the license of a BSD derivation may be easy to accept, > in the case that the application in business with one, > although it is natural to use FDL and the licenses of > other GNU derivations, if I think with the position > that says FSG. > So, comment your idea about license of these documents, please. As far as licensing goes, BSD or LGPL is suitable for the code, but simple copyright (with FSG as the copyright holder) should be used for the spec since you don't want to have others modify the spec once it has been accepted. That is, specifications/standards documents are normally controlled by the organization that produced them... -- ______________________________________________________________________ Michael Sweet, Easy Software Products mike at easysw dot com Internet Printing and Document Software http://www.easysw.com