Mailing-List: contact cygwin-apps-help AT sourceware DOT cygnus DOT com; run by ezmlm
Sender: cygwin-apps-owner AT sourceware DOT cygnus DOT com
List-Subscribe: <mailto:cygwin-apps-subscribe AT sources DOT redhat DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin-apps/>
List-Post: <mailto:cygwin-apps AT sources DOT redhat DOT com>
List-Help: <mailto:cygwin-apps-help AT sources DOT redhat DOT com>, <http://sources.redhat.com/lists.html#faqs>
Delivered-To: mailing list cygwin-apps AT sources DOT redhat DOT com
To: Jeff Sturm <jsturm AT one-point DOT com>
Cc: Anthony Green <green AT redhat DOT com>,
        "Billinghurst, David (CRTS)" <David DOT Billinghurst AT riotinto DOT com>,
        "'Robert Collins'" <robert DOT collins AT itdomain DOT com DOT au>, java AT gcc DOT gnu DOT org,
        cygwin-apps AT sources DOT redhat DOT com
Subject: Re: gcj && libjava for cygwin
References: <Pine DOT LNX DOT 4 DOT 10 DOT 10104181511570 DOT 14265-100000 AT one-point DOT bright DOT net>
X-Zippy:  Actually, what I'd like is a little toy spaceship!!
X-Attribution:  Tom
Reply-To: tromey AT redhat DOT com
From: Tom Tromey <tromey AT redhat DOT com>
Date: 18 Apr 2001 22:23:08 -0600
In-Reply-To: Jeff Sturm's message of "Wed, 18 Apr 2001 15:21:05 -0400 (EDT)"
Message-ID: <87hezlcyc3.fsf@creche.redhat.com>
Lines: 25
X-Mailer: Gnus v5.7/Emacs 20.5

>>>>> "Jeff" == Jeff Sturm <jsturm AT one-point DOT com> writes:

>> JNI_OnLoad is a weak symbol -- see natFirstThread.cc.  Shared libraries of
>> JNI code define it. Do weak symbols not work on cygwin?

Jeff> They don't.  What is the proper workaround?  Define a JNI_OnLoad
Jeff> stub in a static archive?

The only reason JNI_OnLoad appears in FirstThread is to support
preloading of JNI libraries.  This is a libtool feature that is used
on seriously losing platforms that we probably don't support anyway.

Preloading might also be used in other situations, but I don't know.
I've never even tried to use it (which actually doesn't mean much, I'm
just saying it for dramatic effect :-)

Anyway, the point is that supporting this code isn't really a
requirement for us.  Defining JNI_OnLoad in libgcj isn't what we want.
It would be better to simply not define it on Windows.

For JNI libraries loaded dynamically, we find JNI_OnLoad via a
different method.  Not using this code in FirstThread won't affect
that scenario.

Tom