gmane.comp.java.classpath.devel

Re: Free Java Meeting at Fosdem - Brussels,Belgium on 7 and 8February 2009

2008-12-01T14:28:06

Cant wait!!! ----- Original Message ----- From: "Onno Kluyt" Sun.COM> To: "Mark Wielaard" klomp.org> Cc: "classpath" gnu.org>; "discuss" openjdk.java.net>; "java" gcc.gnu.org>; kaffe.org> Sent: Monday, December 01, 2008 2:01 PM Subject: Re: Free Java Meeting at Fosdem - Brussels, Belgium on 7 and 8 February 2009 -------------------------------------------------------------------------------- Internal Virus Database is out of date. Checked by AVG - http://www.avg.com Version: 8.0.175 / Virus Database: 270.9.4/1789 - Release Date: 14/11/2008 19:32

Possible bug in NIO DatagramChannel

Mark Gertsvolf — 2008-12-01T02:23:29

I have been fighting with a small piece of code that fails with Cacao and JamVM with GNU classlib 0.97.2, but works as expected with SUN and IBM JVMs as well as if compiled into native using GCJ. I managed to nail it down to the following small piece of code (exception handling omitted): Selector selector = Selector.open(); DatagramChannel dc = DatagramChannel.open(); dc.configureBlocking(false); dc.socket().bind(socketAddress); dc.register(selector, SelectionKey.OP_READ); selector.select(); Selector is not interrupted when datagrams are sent to the address specified. strace shows that two sockets are opened and bind is called on the second one, but select is executed with the first socket. In the code I am noticing that gnu.java.nio.DatagramChannelImpl and gnu.java.net.PlainDatagramSocketImpl both have private channel members of class VMChannel and channel.initSocket(false) is called for both causing two sockets to be created. First during DatagramChannel.open() DatagramChannelImpl constructor does cha

[Help] SIGSEGV on 'g_ascii_table' ;can't resolve symbol 'g_type_init' ; Bus Error

Donald Gim — 2008-11-22T18:44:49

Hello, I'm working on porting 'jamvm+classpath+gtk+directfb'. (HOST : mipsel-linux, BUILD : i386-linux) The gtk-demo seems to work well for now. But when I run the gnu.classpath.examples.awt.Demo, I am running into some troubles. The access to 'g_ascii_table' causes SIGSEGV. The g_ascii_table is a global pointer from Glib. But it doesn't make any trouble on the gtk-demo. I forced the awt Demo to proceed without the access to that variable. And I got 'Bus Error'. It's from the Java math class. (Math.exp, Math.log) The class is used to make the colour space conversion matrix. Anyway, the Bus Error also occurred on the simple test program, which uses just the Math class and System.out.format(). And when I run the awt Demo without unzipping examples.zip, I see the Bus Error. I heard that it's because my target doesn't have enough memory. Is that right? My target can malloc around 88MB at the maximum. I tried to adjust -Xms/mx/ss options of Jamvm by rule of thumb. I tried the ecj and javac-1.7.0-icedtea. And

Re: trouble w/ synchronized close method in PushbackInputStream

Robert Dodier — 2008-11-10T17:05:28

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37983 Well, this is what I did to get it to work. $ cat PushbackInputStream.java-deadlock-patch --- PushbackInputStream.java-original 2008-10-31 11:41:41.000000000 -0600 +++ PushbackInputStream.java 2008-10-31 16:24:25.000000000 -0600 < at >< at > -131,7 +131,7 < at >< at > * * < at >exception IOException If an error occurs. */ - public synchronized void close() throws IOException + public void close() throws IOException { buf = null; super.close(); FWIW & HTH. Robert Dodier

Re: [GCJ] Performance of GUI applications on embedded systems

Mark Wielaard — 2008-11-09T10:10:57

Hi Andrew, On Sun, 2008-11-09 at 00:24 +0000, Andrew John Hughes wrote: I cannot replicate it with jamvm/classpath, so I assume it is gcj only. The erroneous gdk_threads_leave () line is also not in the classpath CVS sources. Cheers, Mark

Re: [GCJ] Performance of GUI applications on embedded systems

Andrew John Hughes — 2008-11-09T00:24:41

2008/11/8 Andrew Haley redhat.com>: So is the bug GCJ only? I got the impression we were also seeing this on Classpath.

Re: qt-peer thread issue

Tom Tromey — 2008-11-08T15:52:11

Donald> I'm working on porting "Jamvm+Classpath+Qt/Embedded. Donald> Do I need any configuration on the Qt(4.4.3) or Donald> Classpath(0.97.2) to avoid this message? My impression is that the Qt peers are not completely robust. AFAIK they have not been worked on in quite some time. So, I would say that chances are good that you've run into a bug. Tom

Re: trouble w/ synchronized close method in PushbackInputStream

Tom Tromey — 2008-11-08T15:50:33

Robert> PushbackInputStream.close is synchronized. Apparently there is a Robert> deadlock: if I cut out the "synchronized" keyword, the program runs Robert> to completion. I am guessing that there is a thread sleeping in read() Robert> or something like that; I didn't investigate further. Could you file this in bugzilla? That way at least it won't get lost. Or, send a patch :-) Tom

qt-peer thread issue

Donald Gim — 2008-11-05T17:24:18

Hi, I'm working on porting "Jamvm+Classpath+Qt/Embedded. Qt examples work well without Jamvm+Classpath. But when I run the 'gnu.classpath.examples.awt.Demo' with the qt-peer, the following debugging message appears. ASSERT failure in QWidget: "Widgets must be created in the GUI thread.", file Qt/src/gui/kernel/qwidget.cpp, line 1050 Do I need any configuration on the Qt(4.4.3) or Classpath(0.97.2) to avoid this message? Thanks in advance.. donald

Problems with Jikes RVM and Ubuntu 8.10

Ian Rogers — 2008-11-06T15:59:18

Hi, has anyone else tried running Classpath on Ubuntu 8.10? When I run DaCapo antlr I'm getting a problem that a test of File.exists and then File.canWrite (apologies if these aren't the exact method names) is failing. I believe the error is the call to exists is giving a false positive and then the call to canWrite fails as the file doesn't exist. One cause may be that the errno value is 0. This affliction may be Jikes RVM specific as we're compiling Classpath with -m32 on Intel 64bit (the platform I'm testing). Before digging too deep I thought I'd ask if any one else had solved the same problem? Thanks, Ian Rogers -- http://www.cs.man.ac.uk/~irogers/

Re: [GCJ] Performance of GUI applications on embedded systems

Mark Wielaard — 2008-11-05T09:44:26

Run it under GDB and post a backtrace please. Thanks, Mark

Re: trouble w/ synchronized close method in PushbackInputStream

Robert Dodier — 2008-11-03T16:57:21

Well, certainly I don't think so either. Fixing this problem safely probably requires a careful analysis of the PushbackInputStream to see what really needs to be synchronized, and by what means. Unfortunately I simply don't have the time to go into it. Agreed, that's a good idea. Well, unfortunately I don't have anything simpler than this: with an instance of the Apache commons.net FTPClient, open the session, do some operations, and then call FTPClient.disconnect. That deadlocks in PushbackInputStream. Sorry I can't be more specific. Robert Dodier

Re: trouble w/ synchronized close method in PushbackInputStream

Mario Torre — 2008-11-01T19:17:53

Il giorno ven, 31/10/2008 alle 12.40 -0600, Robert Dodier ha scritto: Hi! I don't think this is a safe way of resolving bugs, maybe in this case it is (but I have to look at the code), but avoid the deadlock just by removing a synchronized block means that the two threads now have a chance to get in the way of each other's data. We should instead avoid the deadlock in the first place. Maybe the problem is caused by some threads doing unrelated things, so in this case having an object to use as semaphore, instead of "this", may be of help. Can you provide an example causing the deadlock? Cheers, Mario

trouble w/ synchronized close method in PushbackInputStream

Robert Dodier — 2008-10-31T18:40:22

Hi, I'm working with the FTP classes in Apache commons-net 1.4.1. I am working with classpath 0.96.1 on Buildroot Linux on a Gumstix (ARM cpu). When my code calls org.apache.commons.net.ftp.FTPClient.disconnect, the program hangs. Further investigation shows that it gets stuck just before calling in.close() in java.io.FilterInputStream. The FilterInputStream.in variable is of the class org.apache.commons.net.io.FromNetASCIIInputStream, which extends java.io.PushbackInputStream without redefining the close() method. The program doesn't enter the PushbackInputStream.close method. PushbackInputStream.close is synchronized. Apparently there is a deadlock: if I cut out the "synchronized" keyword, the program runs to completion. I am guessing that there is a thread sleeping in read() or something like that; I didn't investigate further. Incidentally when the same program is run w/ Sun's JRE (java 1.6), it runs to completion. At this point I guess I am inclined to plead for trying to reduce the use of synchron

Re: Inquiry On Contributions From Java Programmer

Andrew Cowie — 2008-10-30T05:31:59

Scott, On Wed, 2008-10-29 at 11:45 -0700, Redefined Horizons wrote: So any Java library that works with C code directly is going to use something called JNI, the "Java Native Interface".* It's what you do to set things up so that C code can accept calls from Java, and, once in C, convert data as it crosses the boundary and/or (rather cumbersomely) access Java objects or invoke Java methods. It's quite comprehensive, but rather arcane. There are good reasons for this, but no matter what you do it represents a steep learning curve. After using it a long while you begin to realize that it's mostly the C equivalent of having doing things Java-side with reflection, but there's nothing to get around the fact that you're now in C and there are a hell of a lot of extra things you have to take care of. This is all germane to your question because a Java library (including a one like the Classpath project which is acting as the "Standard" or core or system library [why oh why we didn't use one of those terms in Ja

Inquiry On Contributions From Java Programmer

Redefined Horizons — 2008-10-29T18:45:59

I've been programming in Java for a few years. I'm interested in gaining some experience in C programming, especially as it relates to Linux. I'm curious if GNU Classpath is implemented in the C programming language, and if there are opportunities to work with C while contributing to Classpath. I've been using Debian/Ubuntu for a couple years, and would be especially interested in working on GNU Classpath support for Linux if that was an area that needed help. Thanks for any info. Scott Huey

Re: problem while compiling classpath-0.92

Mark Wielaard — 2008-10-27T10:16:16

No, attach a debugger to the jikes process to see what it is doing while it is running. Cheers, Mark

Re: problem while compiling classpath-0.92

yinjia jia — 2008-10-27T09:48:55

thank you Mark ,do you mean that I should execute "make --debug " to debug the Makefile? this command give information as below : [sbox-gtk: ~/classpath-0.92] > make --debug GNU Make 3.81 Copyright (C) 2006 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. This program built for i686-pc-linux-gnu Reading makefiles... Updating goal targets.... File `all' does not exist. File `all-recursive' does not exist. Must remake target `all-recursive'. Making all in lib GNU Make 3.81 Copyright (C) 2006 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. This program built for i686-pc-linux-gnu Reading makefiles... make[1]: Entering directory `/home/root/classpath-0.92/lib' true top_builddir=.. top_srcdir=.. /bin/sh ./gen-classlist.sh standard Adding java

Re: problem while compiling classpath-0.92

Mark Wielaard — 2008-10-27T09:22:09

Hi yinjia jia, On Mon, 2008-10-27 at 15:23 +0800, yinjia jia wrote: It sounds like a bug in jikes, you might want to attach a debugger to see what it is doing. Also GNU Classpath 0.92 is pretty old. You might want to upgrade to 0.97.2 and try a different byte code compiler (gcj/ecj). Cheers, Mark

problem while compiling classpath-0.92

yinjia jia — 2008-10-27T07:23:03

hello everyone, I am an new commer to classpath and I try to cross-compile classpath. Now I am confused about the compiling process. I successfully configure the classpath with " ./configure --host=arm-linux --with-jikes --enable-jni --disable-plugin", but when I execute make ,it give the fellow information Making all in lib make[1]: Entering directory `/home/root/classpath-0.92/lib' mkdir -p ../gnu/java/locale ../scripts/generate-locale-list.sh > ../gnu/java/locale/LocaleData.java true top_builddir=.. top_srcdir=.. /bin/sh ./gen-classlist.sh standard Adding java source files from srcdir '..'. Adding java source files from VM directory ../vm/reference /usr/local/bin/jikes +Pno-switchcheck +Pno-shadow +F -encoding UTF-8 -bootclasspath '' -extdirs '' -sourcepath '' --classpath ../vm/reference:..:../external/w3c_dom:../external/sax:../external/relaxngDatatype:.:: -d . < at >classes I wait a long time, but It hang in this stage and not move to next. anybody know what is the reasson? or some suggestions? than

Re: GTK+ peers on DirectFB (instead of Xorg)

ffileppo — 2008-10-20T12:05:37

Hi Robert, thanks for your input. I'm a newbie of gtk peers, but I would like to investigate this subject. Would you spare any starting point or hint about this matter? Thank you, Francesco