From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from [2a02:8b8:656::164] (helo=bar.sig21.net) by merlin.infradead.org with esmtps (Exim 4.76 #1 (Red Hat Linux)) id 1TSshZ-0004Qs-9p for barebox@lists.infradead.org; Mon, 29 Oct 2012 17:00:38 +0000 Date: Mon, 29 Oct 2012 18:00:10 +0100 From: Johannes Stezenbach Message-ID: <20121029170010.GA3257@sig21.net> References: <20121022160341.GF21588@game.jcrosoft.org> <1350922048-25699-1-git-send-email-plagnioj@jcrosoft.com> <20121029085055.GV24458@pengutronix.de> <20121029095604.GA31639@sig21.net> <20121029160513.GA29599@game.jcrosoft.org> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <20121029160513.GA29599@game.jcrosoft.org> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: barebox-bounces@lists.infradead.org Errors-To: barebox-bounces+u.kleine-koenig=pengutronix.de@lists.infradead.org Subject: Re: [PATCH 1/3] command: introduce abi version To: Jean-Christophe PLAGNIOL-VILLARD Cc: barebox@lists.infradead.org On Mon, Oct 29, 2012 at 05:05:13PM +0100, Jean-Christophe PLAGNIOL-VILLARD wrote: > On 10:56 Mon 29 Oct , Johannes Stezenbach wrote: > > > > I wonder how this will work in practice. If I use a simple > > /env/bin/init script and someone makes changes to a command > > which isn't used by my /env/bin/init, will it still cause > > my environment to be detected as incompatible? > > > > Maybe it would be a good idea to give the user control > > over when to change COMMAND_ABI_VERSION by putting it into menuconfig? > > > > Either way I guess it means the default environment's init script > > would need to implement automatic update to not lose important settings. > here you loose nothing we just load defaulenv then you load the old one if you > one the code does not save the defaultenv automatlically Well, the whole point of having an environment is that the user can save a customized version. So by definition falling back to the default means you are going to annoy seme users (even though it's usually not difficult to restore the customizations). > > Bottom line: It's much better to not make incompatible ABI changes _ever_. > I disagree here sometimes we do need as the ABI is really not convinian or > POSIX or like this linux one Well, if an ABI or API is fundamentally broken it might be a good idea to fix it. But it should be well justified. And COMMAND_ABI_VERSION seems to make the issue worse and not better, by forcing an incompatibility where there actually might be none. Johannes _______________________________________________ barebox mailing list barebox@lists.infradead.org http://lists.infradead.org/mailman/listinfo/barebox