From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Fri, 24 Aug 2001 12:04:04 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Fri, 24 Aug 2001 12:03:53 -0400 Received: from smtp3.cern.ch ([137.138.131.164]:41937 "EHLO smtp3.cern.ch") by vger.kernel.org with ESMTP id ; Fri, 24 Aug 2001 12:03:37 -0400 To: Tom Rini Cc: linux-kernel@vger.kernel.org Subject: Re: Will 2.6 require Python for any configuration ? (CML2) In-Reply-To: <20010823143443.F14302@cpe-24-221-152-185.az.sprintbbd.net> <20010824083728.J14302@cpe-24-221-152-185.az.sprintbbd.net> <20010824085052.K14302@cpe-24-221-152-185.az.sprintbbd.net> From: Jes Sorensen Date: 24 Aug 2001 18:03:44 +0200 In-Reply-To: Tom Rini's message of "Fri, 24 Aug 2001 08:50:52 -0700" Message-ID: User-Agent: Gnus/5.070096 (Pterodactyl Gnus v0.96) Emacs/20.4 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org >>>>> "Tom" == Tom Rini writes: Tom> On Fri, Aug 24, 2001 at 05:42:47PM +0200, Jes Sorensen wrote: >> The real problems with Perl is that it exercises almost all of your >> libc, uses floating point math, dlopen() and a lot of other >> funnies. Successfully running Perl's test suite is a very good >> indicator for the completeness of your libc. On the other hand gcc >> and the development toolchain are remarkably easy to accomodate on >> that front. Tom> We're getting someplace now. If you question the ability of the Tom> platform to make reliable tools such as perl, why do you think Tom> you'll have good binutils and gcc on the platform? Maybe because I have ''been there done that''! This is exactly how things were when we brought up the ia64 port. Tom> If your Tom> platform doesn't have dlopen() working then yes, python2 will Tom> probably be pissed off. I conceeded this last time too I think. Tom> If you're trying to bring up a platform compiling a kernel is a Tom> good test of having lots of things working. With 2.6 it'll mean Tom> one more thing is at least somewhat working, dlopen(). Do you have any idea how dlopen() actually works? It requires you to have dynamic loading working, that means shared libraries ... please do us a favor and go read the code in glibc in elf/*.c and sysdep//dl-machine.h (plus a couple of other files). Implementing shared library support is ''very interesting'' but it's by far the first thing you do on a new system. That aside, I am not aware whether Python2 actually requires dlopen to work well enough to run CML2. Tom> Or you Tom> can go and make a CML2 interp in C. Or sh. I'm quite happy Tom> cross compiling stuff until things get a bit farther along on a Tom> brand spanking new platform. Well I can tell you that most people would like to compile their kernels on the host even BEFORE they get shared libraries going. Ok, I think I have spent enough time on this discussion now, back to hacking. Jes