Re: [kaffe] problems with kaffe-1.1.2 + ant-1.5.4
El jue, 30-10-2003 a las 18:45, Dalibor Topic escribió: Hola Juan, [...] BTW, you can also use build.compiler=kjc to load kjc in the same VM where ant is running, if you expect performance gains from that. Thanks: build.compiler=kjc works fine [...] You should give us the invocation line of kjc, too. Probably ant messes up an option for kjc, and doesn't use the proper option to call it. That's an issue that using build.compiler=kjc should solve. By echoing call to javac I obtain: . [javac] javac called as: /opt/usr/java/kaffe-1.1.2/bin/javac -d /home/jantonio/public_html/work/test/build -classpath /home/jantonio/public_html/work/test/build:/opt/usr/java/junit3.8.1/junit.jar:/opt/usr/java/junit3.8.1/junit/tests/runner/test.jar:/home/jantonio:/home/jantonio/public_html/work/test:/opt/usr/java/apache-ant-1.5.4/lib/xml-apis.jar:/opt/usr/java/apache-ant-1.5.4/lib/xercesImpl.jar:/opt/usr/java/apache-ant-1.5.4/lib/optional.jar:/opt/usr/java/apache-ant-1.5.4/lib/ant.jar:/ -sourcepath /home/jantonio/public_html/work/test/src -g:none -verbose /home/jantonio/public_html/work/test/src/test/Alumno.java /home/jantonio/public_html/work/test/src/test/Cuenta.java [javac] Kjc: invalid option -- : BUILD FAILED file:/home/jantonio/public_html/work/test/build.xml:33: Compile failed; see the compiler error output for details. ... This is correct, so I suspect that something fails when forking ant to a new process to call the compiler. I'll try to investigate deeper, Anyway I solved my problem :-) Thanks a lot -- Juan Antonio Martinez [EMAIL PROTECTED] Dpto Ingenieria Telematica signature.asc Description: Esta parte del mensaje =?ISO-8859-1?Q?est=E1?= firmada digitalmente
Re: [kaffe] problems with kaffe-1.1.2 + ant-1.5.4
Hi Juan, Juan Antonio Martinez wrote: El jue, 30-10-2003 a las 18:45, Dalibor Topic escribió: [...] You should give us the invocation line of kjc, too. Probably ant messes up an option for kjc, and doesn't use the proper option to call it. That's an issue that using build.compiler=kjc should solve. By echoing call to javac I obtain: . [javac] javac called as: /opt/usr/java/kaffe-1.1.2/bin/javac -d /home/jantonio/public_html/work/test/build -classpath /home/jantonio/public_html/work/test/build:/opt/usr/java/junit3.8.1/junit.jar:/opt/usr/java/junit3.8.1/junit/tests/runner/test.jar:/home/jantonio:/home/jantonio/public_html/work/test:/opt/usr/java/apache-ant-1.5.4/lib/xml-apis.jar:/opt/usr/java/apache-ant-1.5.4/lib/xercesImpl.jar:/opt/usr/java/apache-ant-1.5.4/lib/optional.jar:/opt/usr/java/apache-ant-1.5.4/lib/ant.jar:/ -sourcepath /home/jantonio/public_html/work/test/src -g:none -verbose /home/jantonio/public_html/work/test/src/test/Alumno.java /home/jantonio/public_html/work/test/src/test/Cuenta.java [javac] Kjc: invalid option -- : BUILD FAILED file:/home/jantonio/public_html/work/test/build.xml:33: Compile failed; see the compiler error output for details. ... This is correct, so I suspect that something fails when forking ant to a new process to call the compiler. I'll try to investigate deeper, Anyway I solved my problem :-) I'm glad that it helped. The problem is that kjc uses a different syntax for command line options than what ant expects, so it ends up confusing kjc. Please use build.compiler=kjc. It tells ant to use the option names required by kjc when it calls kjc. You should also file a bug against ant's extJavac compiler binding: it should try to figure out what the external compiler is, and use the appropriate command line option conventions. cheers, dalibor topic ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Weird warnings on powerpc-macosx
Hallo Michael, Michael Franz wrote: Dalibor, Does this mean that Kaffe is working again on Mac OS X? I tried earlier this week and still had build issues. Unfortunatley, is still has one big build issue: it crashes during the compilation of the class library. It receives a signal EXC_BAD_ACCESS in kaffe/kaffevm/kaffe.def on line 2707. The backtrace looks reasonable to me... I'll try digging deeper intop this, but I only ocassionally have access to a powerpc-darwin machine :( cheers ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Weird warnings on powerpc-macosx
Dalibor, I had spent some time on this before, but I had to give up, when I had more time, it seemed like a lot had changed in Kaffe and I didn't have time to relearn the structure. Unfortunatley, is still has one big build issue: it crashes during the compilation of the class library. It receives a signal EXC_BAD_ACCESS in kaffe/kaffevm/kaffe.def on line 2707. The backtrace looks reasonable to me... I'll try digging deeper intop this, Much appreciated! but I only ocassionally have access to a powerpc-darwin machine :( Do you have access to the compile farm at SourceForge? Michael __ Do you Yahoo!? Exclusive Video Premiere - Britney Spears http://launch.yahoo.com/promos/britneyspears/ ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
[kaffe] Compilation failed
Hi all, The compilation of the current Kaffe CVS version because the file gnu/java/net/content/text/plain.java does not exist in Kaffe. This file is declared in Makefile.am of kaffe/libraries/javalib Best regards, Nicolas ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
[kaffe] GC Error while running simplescalar with kaffe
Hi, when i run some of the Spec JVM98 benchmarks with kaffe (July 2002 release) i get this error mem/gc-incremental.c:1064: gcFree: Assertion `!!!Attempt to explicitly free nonfixed object' failed. Does this mean that the GC is getting invoked? has this been a bug in the release? Thanks archana On Tue, 21 Oct 2003, Jim Pick wrote: On Tue, 21 Oct 2003 00:00:15 -0700 (PDT) archana [EMAIL PROTECTED] wrote: Hi, if one wants to simulate the kaffe interpreter with the gc running say using tools like simplescalar, how can one do it? I did a quick Google search, and I found this: http://www.cae.wisc.edu/~mxu/SimpleScalar_Kaffe_Mini_Howto.shtml I'm not sure I understand what he's actually doing there. the gc runs on a separate thread. is it possible to get the effect by making some changes to the code etc.. there was one posting in this mail list ref: Jun 20 2001, sub: Can/How I disable threading? which i think slightly hinted at the changes that could be done to the code to support threads. Can someone please elaborate on this? When Kaffe starts up, one of the first things it does is create separate threads for the garbage collector and finalizer. I think the modification was to not create those threads, and then run Kaffe with a large enough heap size so that there was never any need to call the garbage collector. If it's useful, I could imagine that we could add a flag which would enable startup with the garbage collector turned off. Patches are welcome. Cheers, - Jim ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
[kaffe] CVS kaffe (kaz): libraries/javalib/java/net/InetAddress.java:
PatchSet 4146 Date: 2003/10/31 15:15:42 Author: kaz Branch: HEAD Tag: (none) Log: 2003-10-31 Ito Kazumitsu [EMAIL PROTECTED] * libraries/javalib/java/net/InetAddress.java: (getAllByName) check whether getHostByName returns IPv6 address. If it is the case we build Inet6Address instead of InetAddress. Originally patched by Guilhem Lavaux on 2003-09-09 but lost when last updated. Members: ChangeLog:1.1738-1.1739 libraries/javalib/java/net/InetAddress.java:1.18-1.19 Index: kaffe/ChangeLog diff -u kaffe/ChangeLog:1.1738 kaffe/ChangeLog:1.1739 --- kaffe/ChangeLog:1.1738 Thu Oct 30 22:05:00 2003 +++ kaffe/ChangeLog Fri Oct 31 15:15:42 2003 @@ -1,3 +1,10 @@ +2003-10-31 Ito Kazumitsu [EMAIL PROTECTED] + + * libraries/javalib/java/net/InetAddress.java: + (getAllByName) check whether getHostByName returns IPv6 address. If it + is the case we build Inet6Address instead of InetAddress. Originally + patched by Guilhem Lavaux on 2003-09-09 but lost when last updated. + 2003-10-30 Dalibor Topic [EMAIL PROTECTED] * config/powerpc/sysdepCallMethod.h: Index: kaffe/libraries/javalib/java/net/InetAddress.java diff -u kaffe/libraries/javalib/java/net/InetAddress.java:1.18 kaffe/libraries/javalib/java/net/InetAddress.java:1.19 --- kaffe/libraries/javalib/java/net/InetAddress.java:1.18 Tue Oct 28 12:29:38 2003 +++ kaffe/libraries/javalib/java/net/InetAddress.java Fri Oct 31 15:15:44 2003 @@ -749,14 +749,18 @@ for (int i = 0; i iplist.length; i++) { -if (iplist[i].length != 4) + if (iplist[i].length != 4 iplist[i].length != 16) throw new UnknownHostException (hostname); // Don't store the hostname in order to force resolution of the // canonical names of these ip's when the user asks for the hostname // But do specify the host alias so if the IP returned won't // reverse lookup we don't throw an exception. -addresses[i] = new InetAddress (iplist[i], null, hostname); + // If the length is equal to 16 this is an IPv6 address. + if (iplist[i].length == 16) + addresses[i] = new Inet6Address (iplist[i], hostname); + else + addresses[i] = new InetAddress (iplist[i], null, hostname); } addToCache (hostname, addresses); ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] CVS kaffe (kaz): libraries/javalib/java/net/InetAddress.java:
Kaffe CVS wrote: PatchSet 4146 Date: 2003/10/31 15:15:42 Author: kaz Branch: HEAD Tag: (none) Log: 2003-10-31 Ito Kazumitsu [EMAIL PROTECTED] * libraries/javalib/java/net/InetAddress.java: (getAllByName) check whether getHostByName returns IPv6 address. If it is the case we build Inet6Address instead of InetAddress. Originally patched by Guilhem Lavaux on 2003-09-09 but lost when last updated. Members: ChangeLog:1.1738-1.1739 libraries/javalib/java/net/InetAddress.java:1.18-1.19 Thanks Ito ! Cheers, Guilhem. ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Classpath beans update.
On Fri, 24 Oct 2003 16:10:52 +0200 Dalibor Topic [EMAIL PROTECTED] wrote: James Simmons wrote: Is it possible to integrate beans from GNU Classpath into Kaffe. I need it for Classpath AWT. Done. Please update from CVS, and good luck with the AWT port. Speaking of the AWT switch to classpath, I was thinking of a slightly different approach: what about having classpath's AWT, kaffe's AWT and pocketlinux AWT source trees somewhere under libraries/javalib and libraries/clib such that kaffe-awt.jar, classpath-awt.jar and pocketlinux-awt.jar are built depending on a compile time switch. Would this be useful? It would blow up the CVS by having some intially redundant code, but that could be phased out, as kaffe's and pockelinux' AWTs merge, and their java side is gradually moved over to classpath. what do you think? I'll all for compile-time switching of AWT implementations. Ideally, it would be nice for other things to be selectable at compile time as well, eg. the threading implementation, the garbage collector, etc. Of course, we'd want to keep the ability for embedded users to be able to compile in only one implementation, so they can save space. As far as the AWT goes, it would be nice to have a standard way of packaging up AWT implementations so they can be installed with any JVM (or, at least, the free ones). eg. Clemens Eisserer packaged up Kaffe's AWT so it could be installed elsewhere, eg. on gcj. http://xawt.sourceforge.net/ There are lots of AWT implementations floating around. Since most of the JVMs support JNI, I don't see why they shouldn't be interchangeable, with a little bit of work. Cheers, - Jim ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
[kaffe] CVS kaffe (dalibor): Fixed bug in interpreter's lcmp
PatchSet 4147 Date: 2003/10/31 20:46:56 Author: dalibor Branch: HEAD Tag: (none) Log: Fixed bug in interpreter's lcmp Members: ChangeLog:1.1739-1.1740 kaffe/kaffevm/intrp/icode.h:1.18-1.19 Index: kaffe/ChangeLog diff -u kaffe/ChangeLog:1.1739 kaffe/ChangeLog:1.1740 --- kaffe/ChangeLog:1.1739 Fri Oct 31 15:15:42 2003 +++ kaffe/ChangeLog Fri Oct 31 20:46:56 2003 @@ -1,3 +1,11 @@ +2003-10-31 Helmer Kraemer [EMAIL PROTECTED] + Dalibor Topic [EMAIL PROTECTED] + + * kaffe/kaffevm/intrp/icode.h + (lcd) Fixed underflow on i686-linux. Use straight +comparisons for longs instead of substracting + one number from the other. + 2003-10-31 Ito Kazumitsu [EMAIL PROTECTED] * libraries/javalib/java/net/InetAddress.java: Index: kaffe/kaffe/kaffevm/intrp/icode.h diff -u kaffe/kaffe/kaffevm/intrp/icode.h:1.18 kaffe/kaffe/kaffevm/intrp/icode.h:1.19 --- kaffe/kaffe/kaffevm/intrp/icode.h:1.18 Tue Sep 2 16:40:58 2003 +++ kaffe/kaffe/kaffevm/intrp/icode.h Fri Oct 31 20:46:57 2003 @@ -28,8 +28,8 @@ #defineashr_long(t, f1, f2)(t)[0].v.tlong = ((int64)(f1)[0].v.tlong) ((f2)[0].v.tint 63) #definelshr_long(t, f1, f2)(t)[0].v.tlong = ((uint64)(f1)[0].v.tlong) ((f2)[0].v.tint 63) -#definelcmp(t, f1, f2) lcc = ((f2)[0].v.tlong) - ((f1)[0].v.tlong); \ - (t)[0].v.tint = (lcc 0 ? -1 : lcc 0 ? 1 : 0) +#definelcmp(t, f1, f2) do { jlong l2 = ((f2)[0].v.tlong); jlong l1 = ((f1)[0].v.tlong); \ + (t)[0].v.tint = ((l1 == l2)? 0 : ((l1 l2) ? -1 : 1)); } while(0); #definecvt_int_long(t, f) (t)[0].v.tlong = (f)[0].v.tint #definecvt_long_int(t, f) (t)[0].v.tint = (f)[0].v.tlong ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
[kaffe] some IPv6-related problem?
Hi, Today's regression tests showed: FAIL: UDPTest.java FAIL: SoTimeout.java Let's look into the case of SoTimeout.java. SoTimeout.fail says: java.lang.IllegalArgumentException: Bad port number: 65536 at java.net.InetSocketAddress.init (InetSocketAddress.java:71) at java.net.ServerSocket.init (ServerSocket.java:175) at java.net.ServerSocket.init (ServerSocket.java:149) at java.net.ServerSocket.init (ServerSocket.java:130) at SoTimeout.main (SoTimeout.java:17) Seeing that tryport of SoTimeout went to 65535, ServerSocket(tryport) must have thrown IOException. My FreeBSD machine is set up so that it uses IPv6, and I am afraid this failure is because libraries/clib/net/PlainSocketImpl.c does not fully support IPv6 addresses. For example, gnu_java_net_PlainSocketImpl_socketCreate always calls KSOCKET(AF_INET, type, 0, fd). ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe