From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 10044 invoked by alias); 23 Nov 2002 01:27:56 -0000 Mailing-List: contact rhug-rhats-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Post: List-Help: , Sender: rhug-rhats-owner@sources.redhat.com Received: (qmail 10036 invoked from network); 23 Nov 2002 01:27:56 -0000 Received: from unknown (HELO primary.maxmind.com) (207.44.162.51) by sources.redhat.com with SMTP; 23 Nov 2002 01:27:56 -0000 Received: from localhost (tjmather@localhost) by primary.maxmind.com (8.11.6/8.11.6) with ESMTP id gAN1WLK05168 for ; Fri, 22 Nov 2002 20:32:21 -0500 Date: Fri, 22 Nov 2002 17:27:00 -0000 From: TJ Mather To: rhug-rhats@sources.redhat.com Subject: Re: OT: Apache POI Native Library using GCJ In-Reply-To: <87fzu1t18m.fsf@fleche.redhat.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-SW-Source: 2002-11/txt/msg00005.txt.bz2 On 16 Nov 2002, Tom Tromey wrote: > TJ> I had to compile FormulaParser.java, POIFSViewEngine.java, and > TJ> HexRead.java to *.class using GCJ, because I got a "bad pc in > TJ> exception_table" if I tried to compile the JDK *.class to *.so > TJ> using GCJ. > > I believe Andrew fixed this bug in the trunk gcj. Yep, the bad pc in exception_table seems to be fixed in GCJ CVS. > TJ> Also note that *.java -> *.class using GCJ doesn't work for some > TJ> *.java files. > > Please submit gcj PRs for this. Especially good would be a small test > case that shows the problem. There are two issues with GCJ 3.2, #8457 (submitted by Martin Aliger), and the "bad pc in exception_table" which has since been fixed. I tried running against GCJ 3.3 20021120, and #8457 was still a problem, and there seems to be a problem with CNI Invocation of Java code from C++ see PR #8685. So the example on the web site seems to be broken on 3.3 20021120: http://gcc.gnu.org/onlinedocs/gcj/Invocation.html -TJ