From mboxrd@z Thu Jan 1 00:00:00 1970 Received: (from weis@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id RAA11907 for caml-red; Wed, 28 Jun 2000 17:46:04 +0200 (MET DST) Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id LAA06042 for ; Wed, 28 Jun 2000 11:45:01 +0200 (MET DST) Received: from enseeiht.enseeiht.fr (enseeiht.enseeiht.fr [147.127.18.144]) by concorde.inria.fr (8.10.0/8.10.0) with ESMTP id e5S9j0X12493 for ; Wed, 28 Jun 2000 11:45:00 +0200 (MET DST) Received: from enseeiht.fr by enseeiht.enseeiht.fr; Wed, 28 Jun 2000 11:44:57 +0200 (MET DST) X-Url: http://www.enseeiht.fr Message-ID: <3959C916.D1A3BDFD@enseeiht.fr> Date: Wed, 28 Jun 2000 11:44:54 +0200 From: David Chemouil Organization: ENSEEIHT - LIMA/IRIT/CNRS X-Mailer: Mozilla 4.7 [en] (X11; I; SunOS 5.6 sun4u) X-Accept-Language: fr, en MIME-Version: 1.0 To: caml-list@inria.fr Subject: convenient features Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Sender: weis@pauillac.inria.fr Hi participating to a former discussion about convenient features which could be added to the Caml system, I'd like to inform you of a few critics, constructive critics of course ;-) 1. One thing that really bothers me is the obligation to put object files in the good order, when linking them. As it is possible to generate the dependency graph (ocamldot does it), wouldn't it be possible for the linker to "flatten" it, in order for it to find alone the good order? It seems to me that it works in C for example. So, one would just have to put necessary object files on the command line, in any order. 'Cause when you have 50 object files, or so, it's really boring to find dependencies "by hand". 2. The second point is minor: it seems to me you only need to put the '-thread' option when your program uses the 'threads.cm[x]a' library. So it must be possible to remove the '-thread' option, and have the compiler guess that the program is multithreaded when it sees 'threads.cm[x]a' on the command line. 3. The last point seems important to me. When you use bytecode threads, marshalling with the Marshal module doesn't work well. Isn't it possible that the compiler automatically replaces it by a working one, like it does with Pervasives? dc -- David Chemouil [mailto:chemouil@enseeiht.fr] [mobile: 06 84 16 26 65] Laboratoire d'informatique et de mathématiques appliquées (IRIT-INPT) "Je vous ai fait trop faibles pour sortir du gouffre, parce que je vous ai fait assez forts pour n'y point tomber" -- Rousseau