From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <weigelt@metux.de>
Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39])
	by yquem.inria.fr (Postfix) with ESMTP id 9BFA1BDD8
	for <caml-list@yquem.inria.fr>; Fri, 26 Aug 2005 14:00:46 +0200 (CEST)
Received: from metux.de (seven.metux.de [193.16.1.1])
	by concorde.inria.fr (8.13.0/8.13.0) with ESMTP id j7QC0iVP009241
	for <caml-list@yquem.inria.fr>; Fri, 26 Aug 2005 14:00:45 +0200
Received: (from weigelt@localhost)
	by metux.de (8.12.10/8.12.10) id j7QC0gQo022193
	for caml-list@yquem.inria.fr; Fri, 26 Aug 2005 14:00:42 +0200
Date: Fri, 26 Aug 2005 14:00:42 +0200
From: Enrico Weigelt <weigelt@metux.de>
To: caml-list@yquem.inria.fr
Subject: crosscompile problem 
Message-ID: <20050826120042.GA9061@nibiru.local>
Reply-To: weigelt@metux.de
Mime-Version: 1.0
Content-Type: multipart/mixed; boundary="oyUTqETQ0mS9luUI"
Content-Disposition: inline
User-Agent: Mutt/1.4.1i
X-j-chkmail-Score: MSGID : 430F046C.000 on concorde : j-chkmail score : X : 0/20 1
X-Miltered: at concorde with ID 430F046C.000 by Joe's j-chkmail (http://j-chkmail.ensmp.fr)!
X-Spam: no; 0.00; ocaml:01 sizeof:01 bypassing:01 distro:01 autoconf:01 config:01 stdout:01 config:01 sed:01 sed:01 longlong:01 doesnt:98 ...:98 cellphone:98 rewrite:01 
X-Attachments: name="buildconf-query" name="buildconf.db" 
X-Spam-Checker-Version: SpamAssassin 3.0.3 (2005-04-27) on yquem.inria.fr
X-Spam-Level: 
X-Spam-Status: No, score=0.0 required=5.0 tests=none autolearn=disabled 
	version=3.0.3


--oyUTqETQ0mS9luUI
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline


Hi folks,

it seems that crosscompiling ocaml is currently impossible. 
The whole ./configure stuff heavily relies on running several 
test programs checking target stuff (ie. sizeof()).
Of course this cannot work on cross targets. 

I started to add some options for bypassing such tests, 
ie. --32bit and --64bit, but this doesnt go far enough and 
also requires distro builders to add special fixes for this 
package on every target.

Since almost all packages have to cope with this problem and
also widely used buildsystems like autoconf also have no clean
way of handling this, I suggest moving away this configuration
from individual packages to some central point - an global 
config database.
We query this database by simply calling some given commandline
with the variable name as parameter. The value is simply printed
out on stdout, without linefeed.

For example to get some type sizes:

    /opt/xcompiler/minibox/bin/buildconf-query host.ansi-c.sizes.long


This can be easily solved by a tiny shellscript and some carefully
maintained text database. (see attachement)


I'll now starting to rewrite the configure script ...


cu
-- 
---------------------------------------------------------------------
 Enrico Weigelt    ==   metux IT service

  phone:     +49 36207 519931         www:       http://www.metux.de/
  fax:       +49 36207 519932         email:     contact@metux.de
  cellphone: +49 174 7066481
---------------------------------------------------------------------
 -- DSL ab 0 Euro. -- statische IP -- UUCP -- Hosting -- Webshops --
---------------------------------------------------------------------

--oyUTqETQ0mS9luUI
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename=buildconf-query

#!/bin/bash

DB=/install/config/systems/jail/etc/buildconf/buildconf.db

if [ ! "$1" ]; then
    echo "$0 <query>";
    exit 1;
fi

QUERY="$1";
OUT=`cat $DB | grep -E "^$QUERY:" | sed -e "s/^$QUERY://;" | sed 's/	/    /g;' | sed "s/^ *//"`
echo -n "$OUT"

--oyUTqETQ0mS9luUI
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="buildconf.db"

### system build configuration stuff for: JAIL (HOMER)

host.cpu.family:		x86
host.cpu.type:			athlon-xp
host.cpu.has-mmx:		yes
host.cpu.has-mmx2:		yes
host.name:			jail
host.ansi-c.sizes.short:	2
host.ansi-c.sizes.long:		4
host.ansi-c.sizes.longlong:	8

--oyUTqETQ0mS9luUI--