From mboxrd@z Thu Jan 1 00:00:00 1970 Received: (from majordomo@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id LAA27291; Mon, 8 Oct 2001 11:17:21 +0200 (MET DST) X-Authentication-Warning: pauillac.inria.fr: majordomo set sender to owner-caml-list@pauillac.inria.fr using -f 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 LAA27314 for ; Mon, 8 Oct 2001 11:17:20 +0200 (MET DST) Received: from comtv.ru (mail.comtv.ru [217.10.32.3]) by concorde.inria.fr (8.11.1/8.10.0) with ESMTP id f989HJv05881 for ; Mon, 8 Oct 2001 11:17:19 +0200 (MET DST) Received: from [10.2.64.72] (HELO oyster2) by comtv.ru (CommuniGate Pro SMTP 3.4.7) with ESMTP id 923645; Mon, 08 Oct 2001 13:17:13 +0400 Date: Mon, 8 Oct 2001 13:18:26 +0400 (MSD) From: malc X-Sender: malc@oyster To: Jeff Henrikson cc: caml-list@inria.fr Subject: Re: [Caml-list] Shared object generation patch feedback In-Reply-To: <002901c14fa0$73f85220$0b01a8c0@mit.edu> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-caml-list@pauillac.inria.fr Precedence: bulk On Sun, 7 Oct 2001, Jeff Henrikson wrote: > > At http://algol.prosalg.no/~malc/scaml you will find a patch against > > OCaml 3.02 and some information suitable for producing shared objects > > on i386 ELF systems. > > Nice! It compiled right out of the box, no unexpected difficulties. > > Is there a way to get a asm runtime in a shared lib? (Either for the purposes of having small executables or for calling from > non-caml) Yes. (ocamlopt -shared ... -ccopt "-L`ocamlc -where` -lasmrun -u caml_main") > > I am having a hard time understanding how this works due to the fact that I don't understand the ocaml calling/symbol naming > convention. I've been reading some assembly intermediary files and nm dumps, but I am still confused and am wondering if there's > some general documentation somewhere. > > Here's an example of something that confuses me. I make a file testlib.ml: > > > open Printf;; > > > > let rec print_int_list x = > > match x with > > [] -> () > > | hd::tl -> printf "%d " hd; (print_int_list tl);; > > > > let rec mymap f l = > > match l with > > [] -> [] > > | hd::tl -> (f hd)::(mymap f tl);; > > and I nm testlib.so and get > > > 000010c0 T Testlib_code_begin > > 0000118a T Testlib_code_end > > 000023f4 D Testlib_data_begin > > 00002424 D Testlib_data_end > > 00001170 T Testlib_entry > > 00002428 D Testlib_frametable > > 00001100 T Testlib_mymap_53 > > 000010c0 T Testlib_print_int_list_49 > > So I wonder: > > 1) where do the numbers 53 and 49 come from, and do I care? For example, if they are arbitrary (as I surmise) and socked away in > testlib.so.cmxa, then don't I get screwed if I try to build an program linked against testlib.so.cmxa and then decide that I want > to change the implementation of testlib? If so, is there a way to define a symbol table just from an mli? (A .so.cmi file or > something?) This command only produces a .cmi: OCaml add those suffixes for variety of reasons, something must be done about it (on the linking stage), if people want -shared to take off. > > ocamlopt -shared -o foo.so foo.mli Naturally. There is nothing to compile, leave alone link here :) > > 2) what do the other entry points mean, eg Testlib_entry, and do I care? Those are module toplevel forms. _frametable, _code[data]_begin[end] are used by GC, equality comparisions, marshaling etc. > > Great work if this actually is heading down the path to real system deployment with ocaml! There is still a lot to be done, to make it really useful. But thanks for encouragment, i needed it. > > > Jeff Henrikson -- mailto:malc@pulsesoft.com ------------------- Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/ To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr