IBM Linux Java updated to jdk1.1.8

1999-09-18 Thread noisebrain


http://www.alphaWorks.ibm.com/tech/linuxjvm


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



borland/inprise jit for blackdown 1.2pre2

1999-09-27 Thread noisebrain


http://www.borland.com/jbuilder/linux/

"
The JBuilder JIT for Linux preview release is based on the
proven JBuilder JIT for Windows that has been shipping for
over three years and provides significant performance
improvements for Java 2 applications on the Linux platform. For
example, compilation is 33% faster when using the JBuilder
JIT for Linux in place of the default JIT included with the
Java 2 Linux JDK

noisebrain --- www.idiom.com/~zilla


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: borland/inprise jit for blackdown 1.2pre2

1999-09-27 Thread noisebrain


If you so choose you can download this jit and run it in place
of the one that comes with 1.2pre2.  The blackdown community
is familiar with two similar replacement jit's - tya and shujit.

On Mon, 27 Sep 1999, Riyad Kalla wrote:

> Sorry to sound arrogant... but how does this effect us? Will this go into
> the blackdown release from now on?
> 
> 
> - Original Message -
> From: noisebrain <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Monday, September 27, 1999 11:58 AM
> Subject: borland/inprise jit for blackdown 1.2pre2
> 
> 
> >
> > http://www.borland.com/jbuilder/linux/
> >
> > "
> > The JBuilder JIT for Linux preview release is based on the
> > proven JBuilder JIT for Windows that has been shipping for
> > over three years and provides significant performance
> > improvements for Java 2 applications on the Linux platform. For
> > example, compilation is 33% faster when using the JBuilder
> > JIT for Linux in place of the default JIT included with the
> > Java 2 Linux JDK
> >
> > noisebrain --- www.idiom.com/~zilla
> >
> >
> > --
> > To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> > with a subject of "unsubscribe". Trouble? Contact
> [EMAIL PROTECTED]
> >
> >
> 
> 


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: borland/inprise jit for blackdown 1.2pre2

1999-09-28 Thread noisebrain


I think that the speed depends on what benchmark you run,
though the new borland jit seems generally faster than the
sunwjit that comes with 1.2pre2.

There's a table of timings of several linux jits versus C
at www.idiom.com/~zilla/Computer/javaCbenchmark.html



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



1.2.2 port update

1999-11-11 Thread noisebrain


I didn't see mention of it on this list, but the status page 
appears to be updated as of 10/30:

"There is still one known problem with SMP systems but
I hope we can do a 1.2.2 (and a 1.1.8) release in the second half of
November. "

also there is a 1.3beta port status page from the same date

"The Blackdown team has been licensed to have access to the JDK-1.3beta
source code. A first pass at a port has been done. Note that Sun
integrated several modifications and improvements made by the Blackdown
team into this version. Unfortunately the integration was done early this
year and so many Linux specific improvements and bug fixes in the
networking, file, and
thread code are missing from 1.3beta. "


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



java vs. C benchmark

1999-11-11 Thread noisebrain


Benchmarks updated for borland's javacomp - this jit does not
compile routines that are only called once, so the benchmark main
now calls the top-level benchmark routine with an iteration of 1,
then calls again with the desired iteration count.



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



html in JLabel, etc: update swing in 1.2prev2?

1999-11-22 Thread noisebrain


I have a project where I want badly to use the htlm-in-component stuff
that is part of Swing1.1.1 (you can label your buttons,fields, etc. with
any html, .e.g., 
String htmlText =
"A JLabel with various text styles"
+ "And a new line!";
JLabel myLabel = new JLabel(htmlText);
)

This will work under 1.1 by supplying the appropriate swing on the
CLASSPATH.  I does not work under 1.2pre-v2 - the swing there is
older.  Is there any way to replace or override the swing that
comes with 1.2pre-v2 ?

(I also want to use the cleaner image access stuff that comes as
part of jdk1.2/java2d, thus 1.1. isn't suitable)

thanks!


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



appreciating sun

1999-12-12 Thread noisebrain

I find some of the sun-bashing both counter-productive and
slightly embarrassing.

Personally, I *want* new versions of the jdk to be released 
in synch with the Windows and Solaris releases, rather than
a year later.  I want Sun to support Linux rather than
think that we're all a bunch of crazed flamers.

Sun made a mistake, but they have apologized now in several places.

The "Sun will never be our friend"/"Sun is an evil corporation"
speak strikes me both as irrelevant and as a bit immature.
Despite open source, I doubt that your local supermarket will
be free anytime soon.  Corporations will continue to exist.

Sun is the creator of Java, and they don't have to give it away to 
anyone.  They certainly don't have to port it to Linux... but they 
chose to put some little bit of energy into this.  The community 
does appreciate Blackdown.  Why resent Sun for helping us?

"We don't need Sun"/use Kaffe/etc.  No, I think we do:  Java wouldn't
have become a standard without Sun's backing, and Java wouldn't
be worth my mindshare if it wasn't a standard - there are other
more advanced/interesting languages out there, but most people
can't afford to re-invent the wheel in an advanced but obscure language.
  Further, Sun is driving Java development, Kaffe isn't.  No one
is preventing the open source community from pushing Java ahead,
or from developing a better-than-java successor, etc.  Historically
though Open Source seems to do well at re-implementing well established
standards.  One of my gripes with Linux is that it has a 70s mindset -
anything that is late 90s (streaming media, videoconferencing, java) 
will be part of your packaged RH cd years after it appears on win/mac.

"Sun doesn't get open source."  Well, Sun is a corporation, but
among corporations they seem to get it more than most.  McNealy(?)
said last week that the software is becoming free, they're releasing
source to Solaris (albeit under their license), etc.

Sun has apologized, so has Inprise, let's accept their apologies
for a change?





--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Blackdown JDK vs Sun/Inprise

1999-12-20 Thread noisebrain


Aren't we forgetting something in this discussion?

Average PC has 64M, you want to write an application that runs
on this PC, your dev environment (JBuilder or whatever)
has, in addition to the application, a compiler, the IDE, a debugger...

...your development environment is probably going to need more than 64m.
This would be as true of a C/C++ ide.

That Java may use more memory than C++ running on the native GUI
is a different issue, and has nothing to do with JBuilder.



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



ibm jdk118 > g++ -O

2000-01-14 Thread noisebrain


I was considering whether we needed to write some image resampling
inner loops in C++, and wrote a little bilinear interpolation 
"mock up" resampling routine.

On a Celeron machine the ibm jdk is actually a hair faster than
the C++ version compiled with -O.  Note that the C++ is not
compiled with any more exotic flags, and quite probably 
is not making use of PII instructions, but nevertheless this
is a pretty satisfying result.  I find it a bit hard to reconcile
with my subjective experience of swing interfaces.

The benchmark ("RESAMPLE") java and C++ source codes are at
www.idiom.com/~zilla/Computer/javaCbenchmark.html.



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: VMs with processor specific code generation

2000-02-03 Thread noisebrain


I think I see indirect evidence of the IBM/intel/linux 118 jvm
using PII instructions - see the RESAMPLE benchmark at
   www.idiom.com/~zilla/Computer/javaCbenchmark.html

This is a synthetic benchmark available in java and C++ versions.
On a PII machine the java code runs faster than the C++ code
that is compiled with g++ -O!

I assume this must be due to the jvm using PII instructions while
the g++ -O assuming only pentium IS without additional flags.




--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



jdb tips?

2000-04-13 Thread noisebrain

Hello,

i'm trying jdb for the first time and am having difficulties.

Using sun/inprise/blackdown 1.2.2(feb version) I run my app with 
-Xdebug -Djava.compiler=NONE -Xbootclasspath:... 
and it prints out e.g.
Agent password=3i5347

The jdb tool doc says that I can then run
jdb -host  -password 
but jdb itself says it does not recognize -host.   Jdb's usage message
says it recognizes 
-connect :=,.
or 
-attach 
but the html doc doesn't describe what  or  are.

Thanks for any help




--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




IBM 1.3 impressions

2000-05-04 Thread noisebrain


- it uses the IBM jit rather than hotspot.  Performs similarly
  to IBM's 1.1.8 on a couple benchmarks (i.e. very good performance,
  though we haven't seen what hotspot can do yet).

- The Swing demo is completely rewritten, worth checking out.
  Java2D demo is modified slightly.

- SWING IS MUCH FASTER.  At last!   Java performance has been
  pretty reasonable for a while now, except swing, which crawled
  in 1.2.  


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




linux java debuggers

2000-05-06 Thread noisebrain


Matt Boersma wrote down his impressions of several debuggers-
thought this might be of general interest.

> Can you give me a recommendation of what debugger(s) to try under
> linux?  I've tried jdb but nothing else so far, and i'd like
> to graduate from inserting prints.

I've used:
JBuilder 3.5 Foundation (free one).  Has a nice integrated debugger,
overall the best IDE in my opinion.

CodeGuide 2.5.  Fast tool for a Pure Java IDE, has a decent integrated
debugger but doesn't have specfic variable watches, so you have to
navigate a tree of objects to find the value you're looking for.

BugSeeker 2b1 is a standalone debugger with a slick UI.  It seems to
work with all the 1.2x and IBM's 1.3 JDKs, but I've had problems getting
it to stop on breakpoints sometimes, seems somewhat random.

Forte for Java is a nice IDE and debugger, but is too resource-intensive
for me (P2-400, 256 MB, RedHat 6.1).

Others I could never get working:
JSwat, JIG, Metamata debug.  And I know there are more.  Here's a link
to some of them:
http://bluemarsh.com/java/jswat/links.html



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




jdk1.3/forte problem

2000-06-11 Thread noisebrain


I tried running the recent jdk1.3+hotspot downloaded from sun
with forte community edition 1.0 v502.  

It has a problem whereby the jmenus pull down but then immediately
disappear, so it is impossible to select any menu items.

Any thoughts or suggestions on what to try? / thanks


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




hotspot server jit?

2000-06-14 Thread noisebrain


The download page for the sun 1.3 beta states that 'both client
and server' jits are bundled.   I don't see the server - 
am I overlooking it?  Perhaps it's intended for the final but
didn't make the beta


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




hardware opengl success

2000-07-04 Thread noisebrain

Slightly off topic, but I was able to get a opengl 
card to work under linux and call it successfully from Java.
This took about 1.5 days of configuring, between Xfree86 4 and
the particular card.  The result is dramatically faster for
textured renders.

Particulars are:
RH6.2,
added Xfree86 4.0.0 
Nvidia Geforce2 chip (Elsa card), using their closed-source driver
(an expert told me this is the current fastest card for my application)
Calling thru the latest GL4Java, from several jdks including the
latest 1.3.

Java Swingset and such still run fine but this card does not
speed up java2d significantly.



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: Debugging Java on Linux with JDK1.2.2

2000-07-19 Thread noisebrain


I used a commercial app called "bugseeker" and had fairly good luck
with it.  It has a free 30(?) day eval period.   Forte (free) also has
a debugger but i have not tried it.  I believe the N.Meyer's Java-Linux
book has a simple debugger as one of its examples.

Even if you don't keep it bugseeker might be a good experience - i learned
some things about java debugging.

For example, in jdb you have to explicitly tell jdb to catch 
uncaught exceptions - otherwise jdb itself quits.

---
j.p.lewis   [EMAIL PROTECTED]
012 345 6789 //www.idiom.com/~zilla



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: Vote of 'No Confidence' in SUNs 'guidance' for Java.

2000-10-29 Thread noisebrain


Nice idea but for industry programmers this would only work
if there was an adequate alternative available.   And if such
were available, there would not be a need for the petition!

The kaffe/gcj/classpath stuff is way behind sun java, and so is not
an alternative.  (Interesting question: is it falling further
behind or is it catching up?)
Python is not even in the same category - it's a nice clean scripting language, 
but it's bytecode interpreted - java has near C++ performance;
this is a factor of 10-50x faster.

Unlike many, I give sun credit for producing a pretty good language
(the best *practical* choice out there at the moment, albeit
some of the libraries are better than others), and for being
slightly open with it: the source *is* available, you *can* change,
they are supporting linux, they are working with apple, they
have incorporated suggestions from ibm, netscape, blackdown, etc.

I hear they could be better at all of these things, but unlike many
people I see Sun's partial efforts as being way better than
microsoft and other companies - and I think we should reward
them for this rather than hinting that they're te devil.

For the sun=microsoft crowd, when do you think Microsoft will
- release source for VB or VC++
- then take the linux port of VB under it's wings and officially keep
  it in-sync with the Win version
- incorporate libraries, design fixes and bugfixes from apple, ibm, and
  you



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: Vote of 'No Confidence' in SUNs 'guidance' for Java.

2000-10-30 Thread noisebrain


On Mon, 30 Oct 2000, [iso-8859-1] Martin Schröder wrote:

> On 2000-10-29 22:26:06 -0800, noisebrain wrote:
> > slightly open with it: the source *is* available, you *can* change,
> 
> Since when? Last time I looked the source was "provided for
> educational value only", changing required an extra license.

This (the license) does not negate the fact that the source is available
and can be changed.




--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: Vote of 'No Confidence' in SUNs 'guidance' for Java.

2000-10-31 Thread noisebrain


I still disagree.  The fact that source is available and can
be changed allowed the blackdown group to port java to linux,
without which we wouldn't be on this list in the first place.
The linux java port is very significant to me and many others.
Yes, it's not the same as GNU, and not as desirable, but for
anyone who says it's the same as microsoft and and others,
I'm still waiting to see the VB source ported to linux.

On Mon, 30 Oct 2000, Chris Abbey wrote:

> At 18:11 10/30/00 -0800, noisebrain wrote:
> >This (the license) does not negate the fact that the source is available
> >and can be changed.
> 
> actually it does. What can you (legally) do with that changed source?
> Nothing. Can you give it to me in binary form? no. Can you give it to me
> in source form? no. Can you give it to me in a diff. No. It might as well
> be the old AT&T unix license. Yeah it's still better than say Windows,
> but it isn't GNU/Linux, it isn't xfree86, it isn't apache, it isn't perl,
> it isn't Open Source; but the more people who will buy into Sun's diluted
> concept the more they and the popular press will forget the real meaning.
> Please don't feed their delusions.
> 
> 
> 
> 
> now the forces of openness
>  have a powerful and
>  unexpected new ally
> http://ibm.com/linux/
> 
> 
> --
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
> 


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: java developer feedback

2000-11-13 Thread noisebrain


I program desktop graphics for a large company (Disney).  We develop in
part on Linux and deploy on SGI, but possibly linux in the future.
Elsewhere in the company there are mac clients with sun servers, etc.

My two issues:

1) FONTS.  The fonts look different on every platform, and they
only look good on windows.  In many cases this destroys the
cross-platform promise of Swing - at best the interfaces look
unprofessional; sometimes we need to have different fonts on 
the different platforms (the font that works on one is clipped by
the textbox on the other).
  Naively I would have thought that SUN could pay a graphic artist
to develop a "metal" font and bundle this with swing.  With so
much of the rest of the font rendering done in java, I might think
that this would give a uniform appearance?
  But I know that I'm ignorant of X/truetype/other font rendering issues.

2) Distribute the jre.  This isn't so much of a company specific
issue, but I would like to see the jre widely distributed so that
the promise of the java "platform" becomes a reality.  Asking users
to download 5megs is asking a lot if they have a modem.   
There's been progress on this recently with caldera/turbolinux/etc., 
but RedHat, by far the biggest linux distrib, does not include
anything other than kaffe.  But, they include every other computer
language on the planet, and they include StarOffice5.1 (before
openoffice) and other non-GPL apps on their extra's disk.  Why can't
they include a jre?


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Q: java-plugin and mozilla/netscape6

2000-11-15 Thread noisebrain


In netscape4, I believe that the java plugin does not
replace the internal jvm, and that web pages need to
have a bit of special html to cause them to use the plugin
instead of the internal jvm.

Is this true in mozilla/netscape6?  Or is the java plugin
"the" java for mozilla?  (& does it work on the linux mozilla
at this point?)

thanks


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




OT: copying binary data with Reader/Writer (rather than Stream)

2000-12-31 Thread noisebrain

Hello,

(forgive me since this is mostly off topic, though the answer
might possibly be specific to the operating system.)

I want to write a little java app to copy a url and its relative links.
The url itself is a text html file; some of the links may be binary
files such as .jpg, .ps.gz, etc.

My question: can you copy a binary file such as a .jpg using a *Reader,
rather than a Stream?  Each byte will then be represented with 
a 16bit char, but will writing with a corresponding Writer preserve the
data?

Generally the top-level url should be read using a Reader, 
but I believe it will be difficult to tell if the links are binary
or html without reading them (and even then... some html does not
start with the  tag.  Thus, the easy/general way to do it would
be to just open every file with a Reader, thus my question.

Any other approaches?

Thanks.



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




address for ibm jdk bugs?

2001-01-07 Thread noisebrain


Hello,

does anyone know an email address at ibm that would be interested
seeing bug reports  in their jdk?  I have one that happens in theirs
but not in blackdown or sun.


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




java not working in mozilla

2001-01-15 Thread noisebrain


I've tried to get java working under mozilla several times without luck.

Currently I'm using mozilla 2001010517 (released this month, I believe
it's called 0.7).  I'm launching it with a script that sets the
environment variable NPX_PLUGIN_PATH to the jre/plugin/i386 directory
of the java.  I've tried both the last blackdown and Sun's jdk1.3.0_01.

When I visit a web page with java it's gives the approximate
message
  'this page has info (application/x-java-vm) that needs a plugin, 
  click OK to download it'

Viewing one of the html files in the sun applet demo directory
does nothing (applet doesn't show up); 
Tasks->java console does nothing that I can see (does not bring up
a window).

I'm sure I'm doing something stupid, sorry for this.





--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: java not working in mozilla

2001-01-15 Thread noisebrain


> Mozilla doesn't understand NPX_PLUGIN_PATH.  You have to create a link

This fixed the problem.  Thank you




--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




Re: JRE as part of Linux

2001-01-25 Thread noisebrain


> What is really needed is a pre-started jvm.  When you start up a java
> process, the jvm will fork, and the child will su to you and proceed as
> normal.  I don't know exactly what the jvm is doing when it is taking
> all that time starting up so I don't know how useful this would be.

I like this solution, though I don't see the details.
My guess is that part of the startup time is just that java
has to uncompress the classes zip file, which is big.  The
scheme above would avoid the uncompress.


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




OT: speed penalty for 'implements serialized'

2001-01-31 Thread noisebrain


Another of those dreaded Off Topic questions:

Is there ever any performance penalty for declaring 
that something implements Serializable?

(sorry for the OT; the quality of the answers on this
list outweigh the politeness factor.  And, you would
probably rather see this question than an on-topic
'how do I get java to run under linux' question!)




--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




when does a method get compiled?

2001-03-16 Thread noisebrain

I have a performance problem that would be explained if the
following were true:

x = new Something()
x.methodcall(); // not compiled yet
x.methodcall(); // second call, compiled now?
y = new int[10];// causes gc, DISCARDS THE COMPILED x ??
x.methodcall(); // compiled version gone interpreted now?

I.e., I'm wondering if a gc can discard a compiled method and
cause it to run interpreted again the next time it is called.
Is this possible?



--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]




vm+class sharing, startup time/Mac OS X

2001-03-19 Thread noisebrain


There was a discussion here recently about sharing
the vm & core classes vs other approaches to reducing startup time.

Apropos of that discussion, the Mac OS X java apparently
does this.   Announcement of its features at

http://developer.apple.com/javaosdn.html






--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]