Hi,
I've been experimenting a little bit with running Mauve tests on top
of Jikes RVM and GNU Classpath. This generally does pretty well, except
that there are massive failures on java.lang.Character.unicode (make check
KEYS=!java. java.lang.Character yields 136460 of 3603803 tests
Thanks for the information. It's useful to know that the test and libraries
are out of synch.It only takes about 20 seconds to run all of the
java.lang.Character.unicode tests on Jikes RVM, so we'll probably just
xfail the ones that are currently failing but go ahead and run the rest of
This is an old chestnut. ;) For some VMs, Double.isNaN() will translate
to
an (expensive) JNI method call. For others it's no big deal. So on the
whole a != a seems like the better bet. Hm, I see Wonka's
java.lang.Math
contains 21 a != a constructs and 2 Double.isNaN(a)'s. :-/
What we
Hi,
I'm putting the finishing touches on a script to run mauve tests
against the CVS head of classpath Jikes RVM. We plan to run this as part
of our nightly regression tests. Is there interest in having these results
periodically (once a week? every day?) posted to the automauve
All of the java.security tests are passing
on Jikes RVM with the cvs head of classpath mauve. We run mauve
tests every night as part of our automated regression testing (results
are posted to the jikesrvm-regression mailing list if anyone is interested).
We also run the Jikes RVM and classpath
Hi,
I'd like to make some minor changes to java.lang.String,
java.lang.Float, and java.lang.Double to make then work with Jikes RVM
slightly better. The desired effect of the changes are:
(1) move implementation of java.lang.String.intern() to another
file so we can provide a
The resulting JikesRVM binary and image
keep seg faulting however on my machine
The most likely explanation for this is that your
glibc is not using the GS register for pthread local state. You can kludge
around this by building Jikes RVM with RVM_FOR_SINGLE_VIRTUAL_PROCESSOR=1.
However, the
Hi,
I
submitted a patch (#1686) for this about two weeks ago. The change
to java.lang.String is actually fairly important for Jikes RVM (our implementation
of String.intern is buggy on classpath until this goes in). Should
I resubmit as two patches to make it easier for someone to apply the
There was some discussion a month or
so ago about making a classpath 0.06 distribution. Is there a chance
this could happen sometime relatively soon? We're thinking about
making the next release of Jikes RVM in 2-3 weeks and it would be great
to be able to get all the hard work done in classpath
Appended is a patch to avoid a NullPointerException
if close is called multiple times. This matches the logic that is
already in java.io.BufferedReader. Could someone please apply it?
Since preparations for classpath 0.06
are underway, I'm increasing the scope of the nightly regression tests
we
Hi Tom,
We've
had security manager checks in the reflection code in Jikes RVM for a while.
Our current workaround for serialization is that classes loaded by
the system classloader are always allowed access. This is probably
too big of a hole, but it does work in practice. The key routine
I've been tracking down a bug using classpath to run JSPs on top of Jikes
RVM and I think the root of the problem is that EncoderUTF8.java is
strictly following the UTF8 encoding scheme instead of the pseudo-UTF8
that JVMs actually need. In particular, the character \u is being
encoded
Because of its m-to-n threading model,
Jikes RVM relies on --enable-portable-native-sync to run AWT code. This
worked in classpath 0.05 but fails in classpath 0.06 with the following
error message:
GThread-ERROR **: The supplied thread
function vector is invalid.
aborting...
Abort
Was this an
That makes sense. I didn't realize
that there was a GTK version switch.
thanks,
--dave
John Leuner [EMAIL PROTECTED]
09/12/2003 10:19 AM
To:
David P Grove/Watson/[EMAIL PROTECTED]
cc:
[EMAIL PROTECTED]
Subject:
Re: --enable-portable-native-sync
Did
Jikes RVM uses a mostly unmodifed classpath.
We don't require users to patch any classpath sources, but there
are currently 12 non-VM classes for which we provide our own implementation.
java.lang.ref:
PhantomReference, Reference, SoftReference, WeakReference
java.lang:
Class, Object,
I am also strongly in favor of putting all VM-specific
native methods in
VM* classes, and all library-specific native methods outside of VM*
classes.
I suspect the notion of a VM-specific native method
vs. a library-specific native method is pretty fuzzy. One example
is VMFloat, most VM's
Julian Dolby knows more about this (maybe
he will jump in), but when you run eclipse on Jikes RVM we've written a
plugin such that the classpath libraries are not only used by eclipse,
but are also used by the applications you develop in eclipse (and Jikes
RVM is used to execute them).
So, you
To:
Andrew Haley [EMAIL PROTECTED]
cc:
David P Grove/Watson/[EMAIL PROTECTED], GNU
Classpath [EMAIL PROTECTED]
Subject:
Re: VMInterface addition: Make native
library namespart
ofVMInterface
David I don't think there is an easy solution
to this as it is unlikely
David that a single
Jikes RVM uses a level of m-to-n threading
and thus relies on the portable locking implementation that Tom mentioned.
In classpath 0.06 this was not the default
and the function vector hadn't been updated to GTK 2.0 (so it didn't work
at all). I just tried with the cvs head of classpath, and it
Hi,
I
tried to run eclipse 3.0M4 on Jikes RVM + classpath CVS head earlier this
week. It died almost immediately because java.nio.channels.FileChannelImpl.tryLock
isn't implemented.
Mark
had mentioned a couple weeks back that he had run eclipse 3.0M4 on gij.
Is this a portion of the
For what it's worth, we (Jikes RVM)
have been getting complaints from users who are trying to browse Jikes
RVM source code using Eclipse that it is somewhat annoying to have tabs
in the source code. If a tab represents 8 spaces, but your spaces
per indentation level is something else (eg 2 or 4)
Hi,
We
thought some people on this list might be interested in knowing that we've
released a new version of Jikes RVM that is using classpath 0.07. If
this kind of posting is not appropriate, let me know and we won't cross
post release announcements in the future. Appended below is the
Hi,
We're
planning on releasing the next version of Jikes RVM in the next 1-2 weeks.
If classpath 0.08 was going to come out soon, we would probably stall
the Jikes RVM release a few days to wait for it. If not, we'll
catch classpath 0.08 on the Jikes RVM release after this one. No
big deal
Jeroen Frijters wrote:
The reference implementation of VMClass will not have any instance
members, but VMs might choose to add instance state. However, after
thinking about it some more, I think it would be better to just add
an
instance member to Class, called vmState (or whatever) of type
Answering Mark's question: Why does
Jikes RVM override 11 non-VMFoo classes?
(1) Native methods: For us, native
methods are (1) lower performance and (2) can't be used early in the VM
booting process. This is the primary reason for java.lang.Object,
java.lang.reflect.Field,
I would be interested in a quick poll of VM implementors
using Classpath:
do you care more about having fewer diffs with stock Classpath
or
modifying optimizing your VM's core classes to eke out optimal
performance?
Both of course ;-)
More seriously, I'm a little cautious when thinking
So,
I
have to admit that I haven't been following this thread in all of its glory
:-), but the JNI spec is quite clear on the sizes of the various primitives.
See the table in section 12.1.1 of the spec. A jbyte is signed
8 bit quantity. A jchar is an unsigned 16 bit quantity, etc. There
are
Development Jikes RVM can take a while
to build, but does buy some performance. Probably not worth it for
classpath development or mauve tests, but if you want to run some apps
it can pay off. I have prototype and development 2.3.2 images sitting
around on my machine, so just for fun.
Jikes RVM 2.3.2 is now available
from the developerworks web site (http://www-124.ibm.com/developerworks/oss/jikesrvm/).
The release notes are appended below. Of particular note is that
of version 2.3.2 it is possible to build Jikes RVM entirely with free software
tools.
*Highlights of this
Sigh.
Building
a development version of Jikes RVM is pretty stressful on the hosting VM.
It's found bugs in more than one product VM in the past. This
looks like one of the nastier forms; the host VM managed to write the RVM.image
file without any obvious error, but the image file is
And if your VM is sufficiently weird
enough (Jikes RVM with m-n threading), then you might actually want to
do something different about files. We used to have our own version
of FileDescriptor and do non-blocking I/O with hooks back into our threading
layer instead of using the default classpath
I'm doing some initial performance comparisons
of classpath 0.08 and classpath 0.09 on Jikes RVM (cvs head). I've
found that the performance of _228_jack (one of the SPECjvm98 benchmarks)
suffers a serve degradation on classpath 0.09. It is twice as slow
(10 seconds vs 5 seconds on my machine
Here's some more data: dynamic
count of JNI functions executed in 1 iteration of jack size 100 on classpath
0.08 and classpath 0.09 with Jikes RVM. Note the massive number of calls
from FileChannelImpl.get_native_fd 2/3 of which are unnecessary. Unless
I hear that someone else has already fixed
Here's a patch. Works with Jikes
RVM and reduces our classpath 0.09 performance degradation on _228_jack
to a slightly more tolerable 5%. Eliot's point is probably a good
one though, this is an unfortunate place to be having to cross JNI boundaries.
Probably the right long term solution for Jikes
IIRC the biggest problem with using JUnit is
that it depends of rather
advanced JVM features; like reflection.
Sorry if this is an ignorant question, but is this
really an issue? I'd always assumed (without actually checking) that
most (all?) of the VMs being used by classpath developers
Archie Cobbs wrote:
Before I go and implement this, does this sound correct to everyone?
I think we desperately need a VMClassLoader.getCallerClassLoader()
method. That way we can simply modify the various load/loadLibrary
methods to call that. It's also much more efficient to implement
Jikes RVM 2.3.3 is now available from the developerworks
web site (http://www-124.ibm.com/developerworks/oss/jikesrvm/).
The release notes are appended below.
-*-outline-*-
These are release notes for Jikes RVM release 2.3.3.
*Sections of this document:
*Highlights
*Other Notes
*Bug Fixes
It depends on what your definition of
free is, but you night want to checkout the DaCapo benchmarks (http://osl-www.cs.umass.edu/DaCapo/gcbm.html).
The main obligation of the license is that if you use the benchmarks you
need to cite them correctly and report the version number of the benchmark
I think you need to understand the point
of the benchmark suite. The whole goal is reproducible science, so
if someone doesn't cite the exact version of the benchmarks, then it isn't
useful (in an academic sense). The license forces that plus proper
academic credit (ie a citation) for the
I should have confined myself to giving
you a pointer to the benchmark suite. Discussion of licensing issues
of non-classpath code is basically off topic spam on this list and I apologize
for starting it. There are better places for these kinds of never
ending discussions.
Jikes RVM version 2.3.5 is now available
for download from sourceforge.net at http://prdownloads.sourceforge.net/jikesrvm/jikesrvm-2.3.5.tar.gz?download.
The Release Notes are appended below.
*Jikes RVM 2.3.5 Release Notes (released on April
8, 2005)
*Highlights
** This release of Jikes RVM
For what it's worth, given this bug Jikes RVM is likely to stick at 0.15
until either 0.16.1 or 0.17 comes out.
--dave
___
Classpath mailing list
Classpath@gnu.org
http://lists.gnu.org/mailman/listinfo/classpath
)
+ return (char[]) value.clone();
+
char[] copy = new char[count];
VMSystem.arraycopy(value, offset, copy, 0, count);
return copy;
[attachment signature.asc deleted by David P Grove/Watson/IBM]
___
Classpath mailing list
Classpath
method. I should
have made it more clear that I was talking about arraycopy in particular,
not native methods in general.
--dave
Archie Cobbs [EMAIL PROTECTED] wrote on 07/12/2005 10:24:10 AM:
David P Grove wrote:
So, I'm having a hard time seeing how this optimization actually
makes
Thanks Mark!
I just tested classpath cvs head against jikes rvm cvs head on SPECjbb2000
and all is happy again.
--dave
Mark Wielaard [EMAIL PROTECTED] wrote on 07/15/2005 08:52:44 AM:
Hi,
On Mon, 2005-07-11 at 07:54 -0400, David P Grove wrote:
This started showing up in regression
yep, that's pretty much the situation with m-n threading in Jikes RVM. For
the blocking operations, we need to replace the entire call with something
else. There might be some cases where the callbacks could be useful, but
they wouldn't be a complete solution for us.
--dave
Guilhem Lavaux
Hi,
Some change(s) to GNU classpath committed in the last 24 hours
caused regression test failures on a number of Jikes RVM tests targeting
floating point. The same cvs checkout of Jikes RVM passes the tests using
classpath 0.20. My guess is that the problem is in the changes related
the patch resolves all the problems; sanity running clean again with
classpath cvs as of an hour ago. thanks,
--dave
Mark Wielaard [EMAIL PROTECTED] wrote on 02/27/2006 07:07:01 AM:
Hi David,
On Thu, 2006-02-23 at 21:29 -0500, David P Grove wrote:
Some change(s) to GNU
The historical reasons why Jikes RVM uses green threads aren't that
relevant to this list, so I'll just direct any one who cares to recent
discussion on the Jikes RVM core list. Personally, I think Jikes RVM
should be moving to native threading mainly to make it easier to run apps
with native
Hi,
I'm getting the following errors when building classpath
0.93-generics for AIX. Anyone seen it before or have advice on how to fix
it? This appears to be in code added since 0.92-generics, which does
build on AIX.
thanks,
--dave
/usr/gnu/bin/gcc -w -O -Wa,-mppc
)
+ return (char[]) value.clone();
+
char[] copy = new char[count];
VMSystem.arraycopy(value, offset, copy, 0, count);
return copy;
[attachment signature.asc deleted by David P Grove/Watson/IBM]
___
Classpath mailing list
Classpath
51 matches
Mail list logo