d to take a look at it
couple of times, and altough no doubt a great VM, it hasn't so far
provided what we need (Standards compliant and nearly complete AWT which
can support Swing on very light GUI library, Xalan/Xerces operation etc.
and all thi
estricted environment use taken more into consideration, or
a statement to the effect that it isn't a design objective for Kaffe.org's
version if that's the case.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
in the default lib? It seems
implemnting at least some of the more trivial IO classes in native code
could yield significant speed-ups for them.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
.println(sbis.read());
} catch (IOException ioe) {
}
}
}
/* Expected Output:
foobar
120
*/
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
ion).
But, well, if we get an usable XML-database based bytecode post-processing
tool that is intuitive and easy to use, I won't complain, but neither do I
volunteer to build one ;) As for text pre-processing, we could use Cpp,
but I'd much rather have something that looks Java-like,
more elegant way to do it, but couldn't come up with one, so
here's the straightforward patch.
Tue Oct 22 15:22:23 EEST 2002 Jukka Santala <[EMAIL PROTECTED]>
* libraries/javalib/kaffe/net/www/protocol/http/
HttpURLConnection.java: Actually use the proxy socket.
I
trivial, but unfortunately this mod goes to Xalan/Xerces, and I'm
almost certain they wouldn't approve of that patch in their releases.
Which is why I'm asking for more frequent Kaffe releases.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
to get them to work
- though I'm working part-time now, and just keeping the HEAD revisions
working together is enough work not to have to worry about backwards
compatibility ;) I'll ask around if it's something we want to do, though,
but I'd suggest Kaffe try to live up to the "Release early, release often"
model.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
ail. If anybody can come up with an implementation that doesn't need
mark()/reset(), I'd be happy. If not, this corrects few exception issues
and at least works, unlike the old code :)
-Jukka Santala
Index: libraries/javal
#x27;s Kaffe Custom Edition didn't support BigInteger up until the
end, as far as I know, so it may be referring to that version. In general
that codebase-split has been source to great confusion, so best not to
believe everything you read a
K_CUT actually work
yet, this class just guarantess they're defined and Kaffe doesn't bowl
over because of that.
And I'll be unlikely to do that work, because this patch is enough to get
our software work with Kaffe ;) (Actually, it's fairly lot more than just
w
TP-site & HTTP gateway as well ;) Altough I think
that doesn't get you the older stuff...
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
ably ought to have it enabled, so maybe there's room to hook it
into some configure script parameters yet.
-Jukka Santala
Index: kaffe/kaffevm/jit3/machine.c
===
RCS file: /cvs/kaffe/kaffe/kaffe/kaffevm/jit3/machine.c,v
retr
are getting a low number of bug reports (< 10 per
> week). They are great when you are getting a lot of
I agree in general, that's alogn the lines that I suggested a rigid patch
check-in policy isn't neccessary. But once 1.0.7 is out, _hopefully_ the
developer (Or at least user/tes
27;d like to see 1) Bug tracking system for
flagging bugs as required for next release, for future release, etc. 2)
Intermediate beta/development releases, 3) Only critical showstopper fixes
to Release Candidates.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
allocating just
5MB, and may well run (or cause something else to run) out of memory on
<=64MB machines when GC could still recover some. So those are good
options to be aware of for embedded use, too.
-Jukka Santala
___
kaffe mailing list
[EMAIL PRO
, for
this case and 1.0.7 release just getting the debug output working seemed
like a more worthwhile goal to me. It seems like rather minor issue
anyhow, but maybe this patch should go to the 1.0.7 branch and a merge
towards PocketLinux/CE on the CVS head.
-Jukka Santala
Index: config/arm/jit-ar
s - such as static builds if VM is
the only real service on the device, and for non-reusable/reused code, and
avoiding double-compression.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
e on your system). Optimizing system
libraries is probably beyond the scope of a Kaffe embedded document, too,
but it might still be worth mentioning that statically linking well
optimized libraries built for the specific platform might be more useful.
Ofcourse, most of those comments aren't in format
On Wed, 5 Jun 2002, Jukka Santala wrote:
> export PATH=/skiff-2.96/local/bin:/skiff-2.96/local/arm-linux/bin:$PATH
> ../kaffe/configure --with-x
I should note I also tried out various parameters from the debug &
profiling categories, as well as static linking, and none of those
paramet
made easier and more intuitive,
and port jit3 for ARM over from PL, though.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
On Fri, 31 May 2002, Jukka Santala wrote:
> I suggest yout ake a look at the jar-file index "spec" of Java, this
> sounds a lot like that, so it'd be probably a good idea to look at
Since I don't think it's actually specified anywhere in any kind of
detail, I'
but native
compilation of some base packages is one thing we're looking at to offset
the common embedded disadvantage in speed.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
is left
wondering why the different way of referring... are they really the same
after all?
So, I'd like to propose trying to accept some sort of style-guidelines for
Kaffe code, for this and other similiar cases. Even if in the end they
just ended up recommending that people try to use unif
licate this brokedness because it affects how
applications run.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
was originally reported in this bug report
that this problem is in "todays build", does this imply it has worked in
some other recent build?
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
er if it
even result sin overall improvement, what with all the cache- and
garbage-collection overhead, and missed opportunities for optimization. On
the other hand, if anybody wants to take a shot at it, why not ;)
-Jukka Santala
___
kaffe mailing lis
is hard enough to trace back bugs to jikes or
Or maybe for release-use only; at least then you know any errors are
likely to have been introduced by the optimizer ;)
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
On Tue, 28 May 2002, Byrne, Eugene wrote:
> How do I unsubscribe?
See the link at the end of every mail posted through the list.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
wledge of
program internals and experience. Without knowing which code-paths are
going to get heavy use, the VM couldn't (shouldn't?) decide on what paths
to inline without just causing extra memory-use. Ofcourse, hopefully
eventually Kaffe would anyway support selective JIT:ing to implement
tter reason, imageUpdate() has also been modified
to fire notify() in this case.
We probably SHOULD also check ABORT in the same manner, but since I
couldn't immediately come up with a test-case to compare this against
other JRE's, I didn't add that in yet.
-Jukka Santala
Index:
primary Klasses.jar gets messed up due to adverse
change, but I'd recommend against making them separately downloadable, as
matching VM (JNI functions) and Klasses.jar versions would be a pain and
probably even larger source of (frivolious) bug reports.
-Jukka Santala
___
d by it). I also looked at coffeine
> Mark 3.0, that one works well with kaffe, but isn't
> open source or free software. Did you find time to
> take at look at a specific benchmarking toolkit?
I checked aroudn a bit, and was tending towards Ashes, but didn't even try
to egt th
On Sun, 26 May 2002, Archie Cobbs wrote:
> Jukka Santala writes:
> > How does this compare to a native method call? I've been under the
> > impression that JNI calls carry significant overhead, so would not
> > neccessarily be efficient for simple functions. On the oth
e functions. On the other hand we have
to take into account that JNI code could be better optimized for the
target platform, but there's probably still something that could be said
about this performance.
Other comments on this kind of refactoring?
-Jukka Santala
s, as
it appears to now work, let's move on to more important things.
(Should we have Bugzilla or equivalent for bug-tracking?)
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
On Fri, 24 May 2002, Jukka Santala wrote:
> Not sure if I forgot this from the original patch, but in any case this is
> missing from the config.in file in Kaffe, preventing compile with
> xprofile:
Another catch: Klasses.jar needs rebuilding, at least the changes in
HttpURLConne
Naturally, autoconf needs to be ran to re-build the configure script, and
the wrapped lines need to be put back together. This comes straight from
JanosVM. Without it, Kaffe creates an empty config/xprofile-md.h file,
which will prevent xprofile from compiling even with tha
ou can run
java or kaffe HelloWorld, then with "unset CLASSPATH".
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
withotu selling your soul, b) Run it on more platforms than any
single other competing VM (afaik).
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
g implementations, because the API JavaDoc makes for a real poor
spec in general. Luckily Kaffe has already done most of this, and the
outcome with AWT programs is very close to that of 1.1. It even
replicates the same incompatibilities to newer API ;) By
ations out right now, and it'll
certainly be a pain to add to whatever JDK you're using at any given time.
(Unless we build all free JDK's capable to use something like Classpath
and abandon all the work put into non-Classpath classlibs, until a
Classpath v2 comes along
to an empty string (or non-number).
-Jukka Santala
Index: libraries/javalib/kaffe/net/www/protocol/http/HttpURLConnection.java
===
RCS file:
/cvs/kaffe/kaffe/libraries/javalib/kaffe/net/www/protocol/http/HttpURLConnection.j
e with. The expected results are from JDK 1.4, and I
notice Kaffe doesn't do as detailed exception reporting, so either the
expected outputs or exception reporting needs to be adjusted. Completely
untested remain inheritance, interfaces and reflection access control.
-Jukka Santala
impor
ew mods for debug /
Xsmiles already, but that runs on framebuffer only.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
ugh I'm not too familiar with this part of the code, and there's no
exceptions etc. available that early to track execution...)
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
up. I configured with
--enable-static* to start off. The only thing that helps is linking with
the profiling C-library directly, but that presents other problems.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
soon. It's probably related to the same
problem preventing xdebugging as well.
While on the topic, could you comment on why the xprof code has been
changed to use gc_malloc with GC_ALLOC_FIXED alloc-strategy instead of
KLMALLOC? Is that something that the common Kaffe
the end of this week,
so I'm having trouble deciding what to dedicate my resources on, but I'll
try to look at the rest of the patch in case it has anything useful for
debugging the code.
-Jukka Santala
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
ableXProfiling':
/home/jsantala/kaffe-build/kaffe/xprof/../../../kaffe/kaffe/xprof/xprofiler.c:243:
undefined reference to `_gmonparam'
/home/jsantala/kaffe-build/kaffe/xprof/../../../kaffe/kaffe/xprof/xprofiler.c:249:
undefined reference to `_gmonparam&
(0.5) appears to exhibit same problem with the recent Kaffe
desktop edition versions. I'll take a closer look if I have time, but as
usual, if this rings a bell to anyone, particularily in regards to the
latest modifications around this year, let me know. (Or fix it;)
-Jukka Santala
_
51 matches
Mail list logo