Hi,
Currently, jre/lib/tools.jar is automatically added in the bootclasspath in
kaffe's startup script. This is problematic for programs that include other
versions of packages contained in tools.jar (for instance gnu.bytecode).
Cheers for 1.1.6,
Daniel
___
Helmer, thanks a lot for your fix!
With the same setup, here is another bug (I have been seeing it for a lot
time, but could not reproduce it reliably until now):
$ cat trace.t
/// PASS bug
int[][] l = [[0,1],[1,2]];
int[] bar = l.map( (T[] x) => x[1]);
You have new mail.
$ /usr/local/src
Hi,
Here is a reproducible GC bug with kaffe HEAD. It's a regression, as it does
not happen with 1.1.5.
This happens when compiling two small programs with the Nice compiler's
testsuite. Therefore, you'll need the Nice compiler, precisely the nice.jar
file. I've reproduced with several vers
I've had problems doing xdebugging, gdb failing at startup. Here is one
session:
sauternes /tmp/kaffe-utf8 KAFFE_DEBUG=gdb
/usr/local/src/kaffe/install/bin/kaffe -Xxdebug -Xmx199m
-Dnice.raw-traces=true -classpath classes nice.tools.testsuite.TestNice testsuite
GNU gdb 6.1-debian
Copyright 2004
Hi,
Here are two annoyances with assertions on kaffe. I don't think either is
critical, but it would be nice to have them fixed eventually.
Here is the testcase:
public class Test
{
public static void main(String[] args) {
assert false;
}
}
1) kaffe always evaluates assertions. I don't se
GNU bytecode has recently been merged in into Kaffe for the cp-tools
tasks in order to support javap task. It seems that this breaks things
for applications that rely on their own version of gnu bytecode.
I guess the right fix is to create a tools.jar, and move our version of
GNU bytecode over the
Hi,
Here is a simple testcase to exercise System.setOut. It looks as if System.out
is not changed at all in kaffe.
import java.io.*;
public class Kaffe
{
public static void main(String[] args)
{
OutputStream stream = new ByteArrayOutputStream();
PrintStream out = new PrintStream(strea
Hi,
Consider the following jasmin bytecode:
.class public Illegal
.super java/lang/Object
;
; standard initializer
.method public ()V
aload_0
invokenonvirtual java/lang/Object/()V
return
.end method
.method public static main([Ljava/lang/String;)V
.limit stack 2
new java/util/LinkedL
Hi,
It seems kaffe is not converting properly from short to int in some
situations. I haven't been able to reproduce this with Java sources, so I just
attach a bytecode class. Dump:
Method name:"main" public static Signature: 28=(java.lang.String[])void
Attribute "Code", length:56, max_stack:2,
Hi,
Here is a simple testcase that fails with kaffe 2005-03-24 (Debian package
PRECVS12):
public class Test
{
public static void main(String[] args)
{
gnu.bytecode.Type.reset();
}
}
package gnu.bytecode;
public abstract class Type {
public static void reset()
{
}
}
$ jikes-classp
Guilhem Lavaux wrote:
Hi Daniel,
I had a few minutes free and checked it in the CVS. Now whenever there
is an uncaught exception happened kaffe-bin will return with 1 instead
of 0.
Cheers,
Guilhem.
Guilhem, thanks a lot for looking into this. However, I tried with the latest
Debian package, whi
Hi list,
When execution terminates because an exception was thrown and never caught,
kaffe still exits with a 0 exit code. It would make sense to have some
non-zero code in that case, to notify that something went wrong. Sun's JVM
exits with 1 in that case.
Cheers,
Daniel
__
Hi,
When using kaffe's jar tool in update mode, I noticed the jar file ends up
with duplicate entries. Then if you update again, thre entries for the same
file, ...
This makes the jar grow indefinitely, and be invalid.
I put up a testcase at:
http://cristal.inria.fr/~bonniot/jarbug.tgz
Run run m
Hi,
I can't get any non-default locale to work with the following program:
import java.util.Date;
import java.text.DateFormat;
public class Bug
{
static DateFormat longDate = DateFormat.getDateTimeInstance();
public static void main(String[] args)
{
System.out.println(longDate.format(new
Hi,
I'm concerned about the result of getHost() on a java.net.URL, when
there is no host part in the URL. Unfortunately the Sun's javadoc is not
clear in this case (while it is in this case for getFile -- empty string
-- and getRef -- null). However, both Sun and gcj/gij agree on returnin
an em
Hi,
I've tried using kaffe to run the testsuite for the Nice[1] compiler
(which involves a fair share of reflection and classloading). It soon
fails, and after some investigation it's due to file descriptors running
out (watching /proc//fd/*, it fails exactly when 1024 are used,
which is the li
Hi,
I've tried using kaffe to run the testsuite for the Nice[1] compiler
(which involves a fair share of reflection and classloading). It soon
fails, and after some investigation it's due to file descriptors running
out (watching /proc//fd/*, it fails exactly when 1024 are used,
which is the limit)
Hi,
I did the investigation on Bug2. It turns out that the whole story is
about getParent() returning null, which is allowed. So Bug2 is not a
bug. (It took me sometime to find this out, because of the stack frame
printed, both at the Java level and inside gdb. The error was actually a
standar
Hi,
I think I am experimenting two Classloader bugs, in Kaffe CVS (built a
few minutes ago). The good thing is that I have pretty small testcases.
Bug.java simply exercises loading from a URLClassLoader. The catch is
that it is fragile on the presence or absence of a '/' at the end of the
URL,
I am verry sorry. I forgot to add new files to libraries/javalib/Makefile.in
when I copied some files from GNU Classpath.
Please get the newest version of Kaffe from CVS and try again.
Thanks a lot. Surprisingly, it seemed to work even without that (it was
the old files in gnu.java.nio that fa
I haven't tried it with jikes yet, but I've just checked in a fix that worked
for me on kjc.
Thanks, it works with jikes too.
So I can now report that kaffe built correctly, and make check reports
no error (only TestNative.java was SKIP'ed, is that normal?).
Furthermore, it could by used to boo
The build also fails with jikes:
Found 11 semantic errors compiling "gnu/java/nio/CharBufferImpl.java":
55. this.backing_buffer = new char [cap];
^^
*** Semantic Error: The field "backing_buffer" in type
"java.nio.CharBuffer" has default access and is not acces
Actually, it might not be that the classes are missing. It is that the
compiler (kjc) finds errors in them (so they probably get ignored after
that). Although I'm not sure, because there are both java.nio.*Impl and
gnu.java.nio.*Impl. There might be two separate issues.
Or could it be a kjc bug?
Hi,
I'm trying to compile the latest version of Kaffe from CVS.
The build of the libraries fails with several errors, like:
java/nio/LongBuffer.java:80: error:Cannot find class "LongBufferImpl"
[JLS 8]
It seems that some (new?) files are missing from the list in the
Makefile for the libraries (
24 matches
Mail list logo