Hi all,
I'd like to know the status of some bugs from Kaffe. Dalibor, if you don't have all the answers, I'll send a mail to [EMAIL PROTECTED] For each bugs, I'd like to know if:
Hi Arnaud,
1� Is it a bug in the Debian package or is it a problem in Kaffe? 2� Is it resolved or does it still exist?
#207998 kaffe: Passing extra JNI directories breaks bootstrap classes
,---- | Version: 1:1.1.1-1 | | When passing additional JNI directories to kaffe using | -Djava.library.path=..., the normal system libraries no longer | seem to get loaded. `----
Bug in kaffe, fixed since
2003-10-06 Dalibor Topic <[EMAIL PROTECTED]>
* libraries/clib/native/System.c:
New system property kaffe.library.path.
(java_lang_System_initProperties) Set java.library.path initally
to empty. Set kaffe.library.path to library path. * libraries/javalib/java/lang/NativeLibrary.java:
(getLibraryNames) Search kaffe.library.path first, and then look
for native libraries in java.library.path. * libraries/javalib/Klasses.jar.bootstrap:
Regenerated.Reported by: Ben Burton <[EMAIL PROTECTED]>
#136148 kaffe's ClassLoader.getSystemResource( "foo/" ) returns nullregard.
,----
| Version: 1:1.0.6-6.1
|
| The problem arises when requesting a directory, thru
| getResource(). Kaffe returns null, while j2sdk1.3.1 returns a
| proper URL. It would appear that kaffe is wrong in this
`----
Fixed in 1.1.0, when class loading was rewritten to match 1.3 behaviour more closely.
cat t.javaimport java.net.URL;
public class t {
public static void main(String[] args) {
URL u = ClassLoader
.getSystemClassLoader()
.getResource(args[0]);
System.out.println(u);
}
}/usr/local/kaffe/bin/kaffe t "./"file:/home/topic//
#159162 sid Kaffe does not support LFS
,---- | Version: 1.0.6 | | Kaffe fails when extending files to more than 2Gb-1 (when | running freenet). This suggests it has been compiled without | large file support. `----
I haven't found AC_SYS_LARGEFILE in kaffe's configure.in, so it's a bug in kaffe. Has not been fixed yet, patches are welcome.
#169229 kaffe: Please provide real manpages for non-kaffe scripts
,---- | Version: 1.0.7-1 | | The manpages for jar and other non-kaffe (non-java) scripts are | all symlinks to kaffe.1.gz, which doesn't provide any | information about them at all. For all practical purposes, | this is the equivalent of an undocumented.7.gz link. Please | provide real manpages for these. `----
Bug in kaffe. Has not been fixed yet. We're gradually reworking the documentation system to DocBook, and want to generate the man pages from that.
#175877 kaffe: Scrollpane has no size; is otherwise brokenhttp://www.scism.sbu.ac.uk/jfl/hci97/morning/scrollpane/ScrollPaneExample.java
,----
| Version: 1:1.0.7-1
|
| A Scrollpane doesn't seem to want to take up any space. An
| example showing the bug is:
|
`----
I assume that it still exists, as AWT hasn't seen a lot development in kaffe recently. I've played a bit with AWT, but in the long range we want to switch over gradually to GNU Classpath's AWT implementation.
#197090 Assertion failure during babel run/build
This bug is vague...
It seems to be a bug in kaffe wrt to loading an applicatin in a jar file produced by gcj 3.3. Uh. Send me the jar file, and I'll look at it
#200434 jni.h and other header files in different directories
,---- | Version: 1:1.0.7-3 | | I have problems trying to compile a file that includes jni.h. | The compiler fails to find the included files. | | The bug has been closed when 1:1.1.1-1 has been uploaded, but | Adam Heath has reopen it (I don't know why). `----
Fixed by:
2003-08-01 Dalibor Topic <[EMAIL PROTECTED]>
Header reorganization and fixes for BerkeleyDB, required for
OpenOffice build. * config/mips/netbsd1/md.c, config/mips/ultrix4/md.c,
include/files.h, include/files.h, include/jni.h,
include/native.h, /kaffe/kaffe/main.c, kaffe/kaffeh/mem.c,
kaffe/kaffeh/support.c, kaffe/kaffevm/code.c,
/kaffe/kaffevm/debug.c, kaffe/kaffevm/exception.c,
kaffe/kaffevm/gtypes.h, kaffe/kaffevm/inflate.c,
kaffe/kaffevm/itypes.c, kaffe/kaffevm/lookup.c,
kaffe/kaffevm/stackTrace.c, kaffe/kaffevm/stats.c,
kaffe/kaffevm/string.c, kaffe/kaffevm/support.c,
kaffe/kaffevm/support.h, kaffe/kaffevm/thread.c,
kaffe/kaffevm/utf8const.c, kaffe/xprof/debugFile.c,
kaffe/xprof/fileSections.c, kaffe/xprof/gmonFile.c,
kaffe/xprof/mangle.c, kaffe/xprof/memorySamples.c,
kaffe/xprof/sectionFile.c, kaffe/xprof/xprofiler.c:
renamed jtypes.h kaffe/jtypes.h. renamed jmalloc.h to
kaffe/jmaloc.h.#210716 jython causes kaffe to fail with assert error
,---- | Version: 1:1.1.1-1 | | > After removing the JNI lines from jython shell script (see | > issue #207998) kaffe dies with kaffe-bin: machine.c:620: | > installMethodCode: Assertion `e->start_pc <= e->end_pc' | > failed. | | Is this releated to #167936? I would imagine one of the | reasons those lines are in the jython script is because of | bugs like #167936. | | >> #167936 Please include /usr/lib/jni in default JNI search | >> path `----
/usr/lib/jni is debian specific. A debian specific patch would be necessary.
#211357 Kaffe doesn't need libffi on PowerPC any longer
,---- | Version: 1.1.1-2 | | Kaffe doesn't need libffi on PowerPC any longer. Arnaud Vandyck | has built Kaffe 1.1.1 without libffi and all tests | passed. Dalibor Topic also thinks that it isn't needed any
| http://www.kaffe.org/pipermail/kaffe/2003-August/043672.html `----
Kaffe doesn't need libffi on powerpc since: 2000-07-26 Edouard G. Parmelan <[EMAIL PROTECTED]>
* libraries/javalib/kjc.jar: upgrade to version kjc-1.4F-egp1: fix
incorrecte integer narrow conversion in method lookup, avoid
compiler failure with += on String and avoid infinite compiler
loop while optimizing test/regression/Preempt.java. * kaffe/kaffevm/classMethod.c (processClass): don't discard native
code of <clinit> if it's throw an exception (avoid assertion
failure in makeMethodActive() with
test/regression/ExceptionInInitialize.java compiled with kjc). * config/powerpc/linux/config.frag: don't force --without-libffi.
* config/Makefile.{am,in}: added powerpc/callmethod_ppc.hcheers, dalibor topic
-- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
