You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Berlin Brown <be...@gmail.com> on 2005/05/10 14:44:12 UTC
Windows support
Hello, I am not a JVM guru like the others, just a long-time Java
advocate and sit-back-and-watch Harmony person. I have a comment, I
know there will be a strong interest in getting an Apache backed
system on Linux/Solaris/Mac environments. It would be nice to see a
strong effort for Windows support. One of the key selling points of
Java is its portability. Having a Windows port would be key. And,
we all love to hate Windows, but it is pretty pervasive. It would be
nice if the Windows support was part of the initial focus and not an
after thought like some of the research Open-Source/JavaVMs that are
out now.
*Reasons*
Don't get me wrong, I am a long-time linux/unix user. Some will argue
that Java is for server-side Linux/Solaris and this is true. But, I
also see that many Java developers tend to love windows even though
they won't admit it.
I think that as more developers are interested in Java, they will
expect Windows support and consider Java as option for development of
Desktop Applications.
Also, I would suggest a port of C-Code Kaffe/GCJ JavaVM code to a
Microsoft backed compiler, possibly Visual C++ 6/7. Or at least
provide build support for the Windows platforrm.
-- Just my 2 cents.
Re: Windows support
Posted by "Geir Magnusson Jr." <ge...@apache.org>.
On May 10, 2005, at 8:44 AM, Berlin Brown wrote:
> Hello, I am not a JVM guru like the others, just a long-time Java
> advocate and sit-back-and-watch Harmony person. I have a comment, I
> know there will be a strong interest in getting an Apache backed
> system on Linux/Solaris/Mac environments. It would be nice to see a
> strong effort for Windows support. One of the key selling points of
> Java is its portability. Having a Windows port would be key. And,
> we all love to hate Windows, but it is pretty pervasive. It would be
> nice if the Windows support was part of the initial focus and not an
> after thought like some of the research Open-Source/JavaVMs that are
> out now.
>
> *Reasons*
>
> Don't get me wrong, I am a long-time linux/unix user. Some will argue
> that Java is for server-side Linux/Solaris and this is true. But, I
> also see that many Java developers tend to love windows even though
> they won't admit it.
>
> I think that as more developers are interested in Java, they will
> expect Windows support and consider Java as option for development of
> Desktop Applications.
>
> Also, I would suggest a port of C-Code Kaffe/GCJ JavaVM code to a
> Microsoft backed compiler, possibly Visual C++ 6/7. Or at least
> provide build support for the Windows platforrm.
>
There's no reason why this won't run on Windows. It would be insane
not to. We want this to be a universal, portable effort. It will
simply come down to where the Harmony-Platform interface is, and we
should be aware that the world is bigger than posix threads and
berkely sockets :)
geir
--
Geir Magnusson Jr +1-203-665-6437
geirm@apache.org
Re: Windows support
Posted by Matthew French <ma...@camary.co.za>.
Berlin Brown said:
> Don't get me wrong, I am a long-time linux/unix user. Some will argue
> that Java is for server-side Linux/Solaris and this is true. But, I
> also see that many Java developers tend to love windows even though
> they won't admit it.
FWIW: I think running on Windows should be an important goal. Although for
a different reason: having the discipline to be properly cross platform
means that porting to more obscure platforms such as mobile phones should
be easier. And hopefully it enforces consistency on the code...
Anyway, the bulk of the work should be in the libraries, and I would
expect that most of this would already be cross platform as they are
implemented in Java.
- Matthew
Re: Windows support
Posted by Berlin Brown <be...@gmail.com>.
Cool, an assignment.
Hehe.
On 5/10/05, Dalibor Topic <ro...@kaffe.org> wrote:
> Berlin Brown wrote:
> > Hello, I am not a JVM guru like the others, just a long-time Java
> > advocate and sit-back-and-watch Harmony person. I have a comment, I
> > know there will be a strong interest in getting an Apache backed
> > system on Linux/Solaris/Mac environments. It would be nice to see a
> > strong effort for Windows support. One of the key selling points of
> > Java is its portability. Having a Windows port would be key. And,
> > we all love to hate Windows, but it is pretty pervasive. It would be
> > nice if the Windows support was part of the initial focus and not an
> > after thought like some of the research Open-Source/JavaVMs that are
> > out now.
>
> Hi Berlin,
>
> I believe that mostly the reason why the current release of Kaffe has
> very little Windows support (just via cygwin) was the lack of big enough
> interest from developers familiar with and working on Windows. That, and
> the lack of some volunteer who cared enough to drive the issue to
> completion.
>
> I'm glad that you are considering volunteering to fix that problem:
> there are Kaffe ports to Cygwin, as well as a mingw32 port, and ports to
> WindowsCE platforms as well as DOS. You do not have to be VM guru to
> fix build problems, and there have even been patches floating around for
> Kaffe to make it build fine with MSVC++. On a side note, integrating
> MSVC++ into Kaffe's (and other autotools based) build systems should be
> fairly easy using cccl m4 macros, I've done this myself in a different
> project to have a single, automated build system.
>
> On the other hand, for all I know IKVM works great on windows already. :)
>
> > Also, I would suggest a port of C-Code Kaffe/GCJ JavaVM code to a
> > Microsoft backed compiler, possibly Visual C++ 6/7. Or at least
> > provide build support for the Windows platforrm.
>
> Get cygwin, get kaffe 1.1.5 tarball, get jikes 1.22 tarball. unpack
> jikes, configure && make && make install it, unpack kaffe, configure it
> with --disable-native-awt (cygwin did not support gtk last time I
> checked) and --disable-sound (neither did it support ALSA or ESD) and
> --enable-pure-java-math (unless you installed GNU MP libs with headers).
> make, and pester me with bug reports :)
>
> The Cygwin patches to 1.1.4 from Gerrit P. Haase have been merged in
> before 1.1.5 came out, but when I tried out 1.1.5 a few weeks ago, I had
> some problems with Cygwin's make utility. On the other hand, cygwin is a
> fast moving target, and chances are the problems are fixed now.
>
> If you are interested in making it happen, hop on the Kaffe mailing
> list, or the irc://irc.freenode.org:#kaffe channel.
>
> cheers,
> dalibor topic
>
Re: Windows support
Posted by Berlin Brown <be...@gmail.com>.
Java by its very nature should be platform independent(to some
degree). I merely ask that one of those platforms should be Windows.
How many open source java libraries have you used on Windows? Or at
least, wouldn't you agree that many others use open source java
libraries. Many of these libraries exist under the apache umbrella.
Just to name a few, hibernate, lucene, jakarta-commons, tomcat, jboss,
eclipse. So, I think a "working" windows java vm would make the
Harmony project very interesting.
On 5/10/05, Kalecser Kurtz <ka...@gmail.com> wrote:
> I fail to see the point on writing open source software for Windows,
> on windows we already have SUN's and IBM's java impl and I think noone
> would care to have an free(from freedom) java impl.
>
>
> On 5/10/05, Dalibor Topic <ro...@kaffe.org> wrote:
> > Berlin Brown wrote:
> > > Hello, I am not a JVM guru like the others, just a long-time Java
> > > advocate and sit-back-and-watch Harmony person. I have a comment, I
> > > know there will be a strong interest in getting an Apache backed
> > > system on Linux/Solaris/Mac environments. It would be nice to see a
> > > strong effort for Windows support. One of the key selling points of
> > > Java is its portability. Having a Windows port would be key. And,
> > > we all love to hate Windows, but it is pretty pervasive. It would be
> > > nice if the Windows support was part of the initial focus and not an
> > > after thought like some of the research Open-Source/JavaVMs that are
> > > out now.
> >
> > Hi Berlin,
> >
> > I believe that mostly the reason why the current release of Kaffe has
> > very little Windows support (just via cygwin) was the lack of big enough
> > interest from developers familiar with and working on Windows. That, and
> > the lack of some volunteer who cared enough to drive the issue to
> > completion.
> >
> > I'm glad that you are considering volunteering to fix that problem:
> > there are Kaffe ports to Cygwin, as well as a mingw32 port, and ports to
> > WindowsCE platforms as well as DOS. You do not have to be VM guru to
> > fix build problems, and there have even been patches floating around for
> > Kaffe to make it build fine with MSVC++. On a side note, integrating
> > MSVC++ into Kaffe's (and other autotools based) build systems should be
> > fairly easy using cccl m4 macros, I've done this myself in a different
> > project to have a single, automated build system.
> >
> > On the other hand, for all I know IKVM works great on windows already. :)
> >
> > > Also, I would suggest a port of C-Code Kaffe/GCJ JavaVM code to a
> > > Microsoft backed compiler, possibly Visual C++ 6/7. Or at least
> > > provide build support for the Windows platforrm.
> >
> > Get cygwin, get kaffe 1.1.5 tarball, get jikes 1.22 tarball. unpack
> > jikes, configure && make && make install it, unpack kaffe, configure it
> > with --disable-native-awt (cygwin did not support gtk last time I
> > checked) and --disable-sound (neither did it support ALSA or ESD) and
> > --enable-pure-java-math (unless you installed GNU MP libs with headers).
> > make, and pester me with bug reports :)
> >
> > The Cygwin patches to 1.1.4 from Gerrit P. Haase have been merged in
> > before 1.1.5 came out, but when I tried out 1.1.5 a few weeks ago, I had
> > some problems with Cygwin's make utility. On the other hand, cygwin is a
> > fast moving target, and chances are the problems are fixed now.
> >
> > If you are interested in making it happen, hop on the Kaffe mailing
> > list, or the irc://irc.freenode.org:#kaffe channel.
> >
> > cheers,
> > dalibor topic
> >
>
Re: Windows support
Posted by Mark Wielaard <ma...@klomp.org>.
Hi,
On Tue, 2005-05-10 at 10:40 -0300, Fernando Lozano wrote:
> It should be noted that GCJ already works pretty well on Windows, you
> can get it as part of MingW, which is a native port of GCC for Win32.
Some packages and instructions available from:
http://www.thisiscool.com/gcc_mingw.htm
See also the Compiling Eclipse applications for Windows with GCJ/MinGW
tutorial http://www.cs.umanitoba.ca/~eclipse/6-Compiling.pdf
And http://www.cs.umanitoba.ca/~eclipse/
And the attached message was just send to the GNU Classpath mailinglist
for PocketPC/Windows users.
Note that I have never used windows so I don't know how well all this
works in practise.
Cheers,
Mark
Re: Windows support
Posted by Fernando Lozano <fe...@lozano.eti.br>.
Hi,
You are completely out of context here. We are not creating an open
webserver -- we already have many of them, best of breed and market
leaders. You can use the Apache Httpd, Apache Tomcat, Jboss, Jonas and
others. Microsoft IIS was long left behind. I know this is difficult to
perceive at some companies, but take a look at www.netcraft.com and see
what are the most used webservers on the net.
We already have our very successfull scripting languages, also market
leaders, like PHP and JSP. Our databases (MySQL, PostgreSQL, Firebird
and others) are not yet the market leaders but they are growing very
fast and getting many former Microsoft or Oracle customers. And all
these are easy to learn, set-up and use. Companies like Google and
Amazon make lots of money using them, vendors like JBoss, MySQL and Red
Hat also have big profits, and there's a very high probability that your
bank uses these also.
But many of the software packages I told you depends on a Java VM to
run. So far the one we use most of the time is not our Java VM, it isn't
a free software Java VM, but a proprietary one made by Sun Microsystem.
This list aims to share the experience of many free software projects
that are creating a Java VM to replace Sun one and so get the result in
a shorter time frame.
So, if you do not develop Java software, you probably are on the wrong
list. :-)
[]s, Fernando Lozano
>hey wow - this newsletterthingy is great. i want to add somelines to give
>some more live to this channelling.
> would are you all talking about. i know java, apache, webtechniques - what
>are you aiming at.
> is it an attempt to create a basically open source websever package, to
>drop mirosoft with their iis from the market or is it a webserver
>communication standard?
> Can you create easyly your own webtechniques ( a new scriptlanguage,
>database modul, services whatever) with it?
> Do i need superextreme highskilllevel programming experiences to use the
>thing?
> who starts the project, it there somebody who is going to make business
>with it (sun?)
> where do most of the people, working for this project come from?
> thanks, so far xunling
>
>
>
Re: Windows support
Posted by kna KnA'hr <xu...@gmail.com>.
hey wow - this newsletterthingy is great. i want to add somelines to give
some more live to this channelling.
would are you all talking about. i know java, apache, webtechniques - what
are you aiming at.
is it an attempt to create a basically open source websever package, to
drop mirosoft with their iis from the market or is it a webserver
communication standard?
Can you create easyly your own webtechniques ( a new scriptlanguage,
database modul, services whatever) with it?
Do i need superextreme highskilllevel programming experiences to use the
thing?
who starts the project, it there somebody who is going to make business
with it (sun?)
where do most of the people, working for this project come from?
thanks, so far xunling
Re: Windows support
Posted by Fernando Lozano <fe...@lozano.eti.br>.
Hi Kalecser,
Free software benifits everyone, and quality free software for Windows
would be a great incentive for people to try also free OSes.
It should be noted that GCJ already works pretty well on Windows, you
can get it as part of MingW, which is a native port of GCC for Win32.
[]s, Fernando Lozano
>I fail to see the point on writing open source software for Windows,
>on windows we already have SUN's and IBM's java impl and I think noone
>would care to have an free(from freedom) java impl.
>
>
>
Re: Windows support
Posted by FaeLLe <mr...@gmail.com>.
Absolutely agreed Windows support is a pure must if we want this to be wide
spread and in use.
On 5/11/05, Wes Wannemacher <te...@gmail.com> wrote:
>
> On 5/10/05, Bob <ci...@earthlink.net> wrote:
> [snip]
> > I would add to this. It's a story we've seen time and time again.
> > First came GCC. Not only was it a free compiler, but it was also the
> > first compiler that ran the SAME on all (then extant) platforms. Over
> > the course of about 5 years, it overtook and superceded all other C
> > compilers (except Microsoft's) for that reason. This is the power of
> > "embrace and extend", turned around to the free software world.
> >
> > Same thing with Mozilla. I use Firefox on my Linux, Mac and PCs. Why?
> > Why don't I use IE or Safari or Konqueror so much? Because Firefox is
> > cross-platform, I can use it the same way wherever I go.
> >
> > Why has Evolution NOT taken over Email even though it had the lead on
> > full-text indexing for AT LEAST 2 years? (This is something Outlook
> > STILL didn't do last time I checked). Because Evolution runs ONLY on
> > Linux.
> >
> > I believe it's essential to get Harmony running on Windows. It will
> > inevitably be different from Sun's Java: the bugs will be different,
> > some features will be missing, and some extra features will be added.
> > If Harmony runs cross-platform, then a Harmony-java developer can be
> > sure a program will run on Linux/Windows/Mac. If not, then the Harmony
> > developer must always drag out the Sun implementation to make sure the
> > app runs on Sun's Java (which will be required for windows). And once
> > you've dragged out Sun's Java, why bother using Harmony at all? No one
> > wants to do QA on their program twice, if they only have to do it once.
> >
>
> I want to play devil's advocate here and mention that Sun's JVM does
> run on a variety of platforms with few if any platform specific quirks
> (in my experience). And, the reality of the matter is that a project
> *will* be QE'd if it is moved to a completely different host platform.
>
> Although, I do agree wholeheartedly that Harmony should run as
> described on as many platforms as possible. I do think however, that
> it should perform better than Sun's JVM. With the exception of
> idealistic OS distributions (Debian, Gentoo), Sun's JVM is likely to
> continue to be packaged. The users will come if there is a compelling
> reason to do so (i.e. easy integration and optimizations for Tomcat).
>
> Just my $.02
>
> -Wes
>
--
www.FaeLLe.com <http://www.FaeLLe.com>
Re: Windows support
Posted by Mika Hirvonen <hi...@gmail.com>.
On 5/11/05, Vitaly Harisov <vi...@harisov.name> wrote:
> On Wed, 11 May 2005 16:06:23 +0300, Ricky Clarkson
> <ri...@gmail.com> wrote:
>
> > Debian's java-package package does package Sun's JVM, but the user has
> > to download it. I think this is more legal issues than idealism.
>
> Gentoo does it in the same way.
JPackage does it as well. They have a .nosrc.rpm that creates normal
binary RPMs from Sun's JDK tarball.
--
Mika Hirvonen <hi...@gmail.com>
Re: Windows support
Posted by Vitaly Harisov <vi...@harisov.name>.
On Wed, 11 May 2005 16:06:23 +0300, Ricky Clarkson
<ri...@gmail.com> wrote:
> Debian's java-package package does package Sun's JVM, but the user has
> to download it. I think this is more legal issues than idealism.
Gentoo does it in the same way.
--
The End
Re: Windows support
Posted by Ricky Clarkson <ri...@gmail.com>.
Debian's java-package package does package Sun's JVM, but the user has
to download it. I think this is more legal issues than idealism.
Take a look at the indemnification clause in the JVM licence, for one.
There are lots of legal raisins to overcome with Sun Java.
> With the exception of
> idealistic OS distributions (Debian, Gentoo), Sun's JVM is likely to
> continue to be packaged.
Re: Windows support
Posted by Wes Wannemacher <te...@gmail.com>.
On 5/10/05, Bob <ci...@earthlink.net> wrote:
[snip]
> I would add to this. It's a story we've seen time and time again.
> First came GCC. Not only was it a free compiler, but it was also the
> first compiler that ran the SAME on all (then extant) platforms. Over
> the course of about 5 years, it overtook and superceded all other C
> compilers (except Microsoft's) for that reason. This is the power of
> "embrace and extend", turned around to the free software world.
>
> Same thing with Mozilla. I use Firefox on my Linux, Mac and PCs. Why?
> Why don't I use IE or Safari or Konqueror so much? Because Firefox is
> cross-platform, I can use it the same way wherever I go.
>
> Why has Evolution NOT taken over Email even though it had the lead on
> full-text indexing for AT LEAST 2 years? (This is something Outlook
> STILL didn't do last time I checked). Because Evolution runs ONLY on
> Linux.
>
> I believe it's essential to get Harmony running on Windows. It will
> inevitably be different from Sun's Java: the bugs will be different,
> some features will be missing, and some extra features will be added.
> If Harmony runs cross-platform, then a Harmony-java developer can be
> sure a program will run on Linux/Windows/Mac. If not, then the Harmony
> developer must always drag out the Sun implementation to make sure the
> app runs on Sun's Java (which will be required for windows). And once
> you've dragged out Sun's Java, why bother using Harmony at all? No one
> wants to do QA on their program twice, if they only have to do it once.
>
I want to play devil's advocate here and mention that Sun's JVM does
run on a variety of platforms with few if any platform specific quirks
(in my experience). And, the reality of the matter is that a project
*will* be QE'd if it is moved to a completely different host platform.
Although, I do agree wholeheartedly that Harmony should run as
described on as many platforms as possible. I do think however, that
it should perform better than Sun's JVM. With the exception of
idealistic OS distributions (Debian, Gentoo), Sun's JVM is likely to
continue to be packaged. The users will come if there is a compelling
reason to do so (i.e. easy integration and optimizations for Tomcat).
Just my $.02
-Wes
Re: Windows support
Posted by Ryan Bloom <rb...@gmail.com>.
Yeah, but that should already be part of APR's goals, so this would
just be a reason for somebody (me) to make it happen.
Ryan
On 5/11/05, Ben Laurie <be...@algroup.co.uk> wrote:
> Ryan Bloom wrote:
> > One option for us to get full cross-platform support would be to write
> > the JVM using APR. This way we won't have to deal with a lot of the
> > hassles of cross-platform development. There are some tricks to
> > working with the pools, but those can be handled, and if we do this
> > right, it might even lead to changes to APR that will improve that
> > library.
>
> I'm pretty sure pools and a JVM are not a good mix. We'd need to ensure
> APR can run pool-less.
>
> --
> http://www.apache-ssl.org/ben.html http://www.thebunker.net/
>
> "There is no limit to what a man can do or how far he can go if he
> doesn't mind who gets the credit." - Robert Woodruff
>
--
Ryan Bloom
rbb@apache.org
rbb@rkbloom.net
rbloom@gmail.com
Re: Windows support
Posted by Garrett Rooney <ro...@electricjellyfish.net>.
Stefano Mazzocchi wrote:
> For the APR-impared between me (me, for example) could you elaborate a
> little more what this means?
Currently APR requires that all memory allocation be done within a
"pool". You don't malloc and free memory directly, you allocate within
the pool, then either clear it (allowing the memory to be reused all in
one shot), or destroy it (passing the memory back to the OS). Pools are
hierarchical, so you can create child pools that are intended to be used
for temporary allocations. The end result is less memory bookkeeping,
but the program tends to use more memory because you're not cleaning it
up ASAP. Pools are also used to manage other resources (network
sockets, file handles, etc) and even arbitrary user defined resources
via callbacks.
This fits very well with the way something like the Apache HTTPD works,
where you might have a bunch of resources that are scoped to the
request, but after the request you can just throw them all away. It
doesn't necessarily fit with the use patterns of other programs, so Ryan
is suggesting that it would be worthwhile to make the use of pools in
APR optional in some way.
-garrett
Re: Windows support
Posted by Stefano Mazzocchi <st...@apache.org>.
Ben Laurie wrote:
> Ryan Bloom wrote:
>
>> One option for us to get full cross-platform support would be to write
>> the JVM using APR. This way we won't have to deal with a lot of the
>> hassles of cross-platform development. There are some tricks to
>> working with the pools, but those can be handled, and if we do this
>> right, it might even lead to changes to APR that will improve that
>> library.
>
>
> I'm pretty sure pools and a JVM are not a good mix. We'd need to ensure
> APR can run pool-less.
For the APR-impared between me (me, for example) could you elaborate a
little more what this means?
--
Stefano.
Re: Windows support
Posted by Ben Laurie <be...@algroup.co.uk>.
Ryan Bloom wrote:
> One option for us to get full cross-platform support would be to write
> the JVM using APR. This way we won't have to deal with a lot of the
> hassles of cross-platform development. There are some tricks to
> working with the pools, but those can be handled, and if we do this
> right, it might even lead to changes to APR that will improve that
> library.
I'm pretty sure pools and a JVM are not a good mix. We'd need to ensure
APR can run pool-less.
--
http://www.apache-ssl.org/ben.html http://www.thebunker.net/
"There is no limit to what a man can do or how far he can go if he
doesn't mind who gets the credit." - Robert Woodruff
Re: Windows support
Posted by Ryan Bloom <rb...@gmail.com>.
One option for us to get full cross-platform support would be to write
the JVM using APR. This way we won't have to deal with a lot of the
hassles of cross-platform development. There are some tricks to
working with the pools, but those can be handled, and if we do this
right, it might even lead to changes to APR that will improve that
library.
Ryan
On 5/10/05, Kalecser Kurtz <ka...@gmail.com> wrote:
> Ok, agreed, Harmony will inevitably have its peculiarites, so harmony
> wouldn't be an option without full windows and apple support. Besides
> that it would create a stronger community, like Stefano said.
>
>
> On 5/10/05, Bob <ci...@earthlink.net> wrote:
> >
> > On May 10, 2005, at 10:26 AM, Stefano Mazzocchi wrote:
> >
> > > Kalecser Kurtz wrote:
> > >> I fail to see the point on writing open source software for Windows,
> > >> on windows we already have SUN's and IBM's java impl and I think noone
> > >> would care to have an free(from freedom) java impl.
> > >
> > > Some shortsighted people said the same when mozilla started ;-)
> >
> > I would add to this. It's a story we've seen time and time again.
> > First came GCC. Not only was it a free compiler, but it was also the
> > first compiler that ran the SAME on all (then extant) platforms. Over
> > the course of about 5 years, it overtook and superceded all other C
> > compilers (except Microsoft's) for that reason. This is the power of
> > "embrace and extend", turned around to the free software world.
> >
> > Same thing with Mozilla. I use Firefox on my Linux, Mac and PCs. Why?
> > Why don't I use IE or Safari or Konqueror so much? Because Firefox is
> > cross-platform, I can use it the same way wherever I go.
> >
> > Why has Evolution NOT taken over Email even though it had the lead on
> > full-text indexing for AT LEAST 2 years? (This is something Outlook
> > STILL didn't do last time I checked). Because Evolution runs ONLY on
> > Linux.
> >
> > I believe it's essential to get Harmony running on Windows. It will
> > inevitably be different from Sun's Java: the bugs will be different,
> > some features will be missing, and some extra features will be added.
> > If Harmony runs cross-platform, then a Harmony-java developer can be
> > sure a program will run on Linux/Windows/Mac. If not, then the Harmony
> > developer must always drag out the Sun implementation to make sure the
> > app runs on Sun's Java (which will be required for windows). And once
> > you've dragged out Sun's Java, why bother using Harmony at all? No one
> > wants to do QA on their program twice, if they only have to do it once.
> >
> >
>
--
Ryan Bloom
rbb@apache.org
rbb@rkbloom.net
rbloom@gmail.com
Re: Windows support
Posted by Kalecser Kurtz <ka...@gmail.com>.
Ok, agreed, Harmony will inevitably have its peculiarites, so harmony
wouldn't be an option without full windows and apple support. Besides
that it would create a stronger community, like Stefano said.
On 5/10/05, Bob <ci...@earthlink.net> wrote:
>
> On May 10, 2005, at 10:26 AM, Stefano Mazzocchi wrote:
>
> > Kalecser Kurtz wrote:
> >> I fail to see the point on writing open source software for Windows,
> >> on windows we already have SUN's and IBM's java impl and I think noone
> >> would care to have an free(from freedom) java impl.
> >
> > Some shortsighted people said the same when mozilla started ;-)
>
> I would add to this. It's a story we've seen time and time again.
> First came GCC. Not only was it a free compiler, but it was also the
> first compiler that ran the SAME on all (then extant) platforms. Over
> the course of about 5 years, it overtook and superceded all other C
> compilers (except Microsoft's) for that reason. This is the power of
> "embrace and extend", turned around to the free software world.
>
> Same thing with Mozilla. I use Firefox on my Linux, Mac and PCs. Why?
> Why don't I use IE or Safari or Konqueror so much? Because Firefox is
> cross-platform, I can use it the same way wherever I go.
>
> Why has Evolution NOT taken over Email even though it had the lead on
> full-text indexing for AT LEAST 2 years? (This is something Outlook
> STILL didn't do last time I checked). Because Evolution runs ONLY on
> Linux.
>
> I believe it's essential to get Harmony running on Windows. It will
> inevitably be different from Sun's Java: the bugs will be different,
> some features will be missing, and some extra features will be added.
> If Harmony runs cross-platform, then a Harmony-java developer can be
> sure a program will run on Linux/Windows/Mac. If not, then the Harmony
> developer must always drag out the Sun implementation to make sure the
> app runs on Sun's Java (which will be required for windows). And once
> you've dragged out Sun's Java, why bother using Harmony at all? No one
> wants to do QA on their program twice, if they only have to do it once.
>
>
Re: Windows support
Posted by Bob <ci...@earthlink.net>.
On May 10, 2005, at 10:26 AM, Stefano Mazzocchi wrote:
> Kalecser Kurtz wrote:
>> I fail to see the point on writing open source software for Windows,
>> on windows we already have SUN's and IBM's java impl and I think noone
>> would care to have an free(from freedom) java impl.
>
> Some shortsighted people said the same when mozilla started ;-)
I would add to this. It's a story we've seen time and time again.
First came GCC. Not only was it a free compiler, but it was also the
first compiler that ran the SAME on all (then extant) platforms. Over
the course of about 5 years, it overtook and superceded all other C
compilers (except Microsoft's) for that reason. This is the power of
"embrace and extend", turned around to the free software world.
Same thing with Mozilla. I use Firefox on my Linux, Mac and PCs. Why?
Why don't I use IE or Safari or Konqueror so much? Because Firefox is
cross-platform, I can use it the same way wherever I go.
Why has Evolution NOT taken over Email even though it had the lead on
full-text indexing for AT LEAST 2 years? (This is something Outlook
STILL didn't do last time I checked). Because Evolution runs ONLY on
Linux.
I believe it's essential to get Harmony running on Windows. It will
inevitably be different from Sun's Java: the bugs will be different,
some features will be missing, and some extra features will be added.
If Harmony runs cross-platform, then a Harmony-java developer can be
sure a program will run on Linux/Windows/Mac. If not, then the Harmony
developer must always drag out the Sun implementation to make sure the
app runs on Sun's Java (which will be required for windows). And once
you've dragged out Sun's Java, why bother using Harmony at all? No one
wants to do QA on their program twice, if they only have to do it once.
Re: Windows support
Posted by Stefano Mazzocchi <st...@apache.org>.
Kalecser Kurtz wrote:
> I fail to see the point on writing open source software for Windows,
> on windows we already have SUN's and IBM's java impl and I think noone
> would care to have an free(from freedom) java impl.
Some shortsighted people said the same when mozilla started ;-)
You a lot of people fail to see is that the world is smarter than XYZ
Corp. (interestingly enough, even Bill Joy said that out loud once).
Some people like 'free software' for its freedom. Some people like 'open
source' for its openness.
I like it because, if done well, it creates communities and communities
(as once Sam said) are your best copyleft *and* your best driving force
*and* your best catalitic engine.
Not only "given enough eyeballs bugs gets fixed", but given enough
modularity and 'harmonic' reuse, people are free to innovate and
experiment, and communities are free to pick up the best solutions.
On the outside, a JVM needs to pass a TCK in order to be certified and
called "java". But on the inside, there is no tall order on how you
should implement the JIT compiler or the GC algorithm, and these are
great places for people to experiment, including native precompilation
and/or OS-level security sandboxing.
Let me tell you, the day a 'free' JVM will run your code faster and as
reliably on Windows, people will switch, just like the use apache httpd
instead of IIS.
None of us wants to get a JVM that is compliant but a sluggish pure
accademic exercise.
We want to rock the java planet and build bridges between all different
communities and pieces, existing and not, already open sourced or
yet-to-be-donated.
We might fail. Yes, this is known and recognized. But at least we won't
feel guilty anymore for not having tried.
--
Stefano.
Re: Windows support
Posted by Matthew French <ma...@camary.co.za>.
Kalecser Kurtz said:
> I fail to see the point on writing open source software for Windows,
> on windows we already have SUN's and IBM's java impl and I think noone
> would care to have an free(from freedom) java impl.
We have a customer writing a parallel computing application in Java.
Multithreaded performance on Windows sucks big time. But there were also
all sorts of performance problems running under Solaris on an 8 processor
Sparc box.
Currently they are using JBoss with a view to having a cluster of many
J2EE servers. (Which is kind of like swatting a fly with a tank.) And to
make it more interesting, it looks like JBoss will be running on Windows
thanks to the complete lack of awareness about Linux in the organisation.
Now if they had an open source JVM that gave the performance they needed,
with the ability to customise the garbage collector (a big source of the
performance problem) then it would be a win for open source, hopefully
some more developers for the project, and one more big step towards
Linux/BSD in this organisation.
Of course, if developers are not inclined to write the Windows port - I
know it does not hold much interest for me - then we cannot make them. But
we can still design for Windows support...
- Matthew
Re: Windows support
Posted by Kalecser Kurtz <ka...@gmail.com>.
I fail to see the point on writing open source software for Windows,
on windows we already have SUN's and IBM's java impl and I think noone
would care to have an free(from freedom) java impl.
On 5/10/05, Dalibor Topic <ro...@kaffe.org> wrote:
> Berlin Brown wrote:
> > Hello, I am not a JVM guru like the others, just a long-time Java
> > advocate and sit-back-and-watch Harmony person. I have a comment, I
> > know there will be a strong interest in getting an Apache backed
> > system on Linux/Solaris/Mac environments. It would be nice to see a
> > strong effort for Windows support. One of the key selling points of
> > Java is its portability. Having a Windows port would be key. And,
> > we all love to hate Windows, but it is pretty pervasive. It would be
> > nice if the Windows support was part of the initial focus and not an
> > after thought like some of the research Open-Source/JavaVMs that are
> > out now.
>
> Hi Berlin,
>
> I believe that mostly the reason why the current release of Kaffe has
> very little Windows support (just via cygwin) was the lack of big enough
> interest from developers familiar with and working on Windows. That, and
> the lack of some volunteer who cared enough to drive the issue to
> completion.
>
> I'm glad that you are considering volunteering to fix that problem:
> there are Kaffe ports to Cygwin, as well as a mingw32 port, and ports to
> WindowsCE platforms as well as DOS. You do not have to be VM guru to
> fix build problems, and there have even been patches floating around for
> Kaffe to make it build fine with MSVC++. On a side note, integrating
> MSVC++ into Kaffe's (and other autotools based) build systems should be
> fairly easy using cccl m4 macros, I've done this myself in a different
> project to have a single, automated build system.
>
> On the other hand, for all I know IKVM works great on windows already. :)
>
> > Also, I would suggest a port of C-Code Kaffe/GCJ JavaVM code to a
> > Microsoft backed compiler, possibly Visual C++ 6/7. Or at least
> > provide build support for the Windows platforrm.
>
> Get cygwin, get kaffe 1.1.5 tarball, get jikes 1.22 tarball. unpack
> jikes, configure && make && make install it, unpack kaffe, configure it
> with --disable-native-awt (cygwin did not support gtk last time I
> checked) and --disable-sound (neither did it support ALSA or ESD) and
> --enable-pure-java-math (unless you installed GNU MP libs with headers).
> make, and pester me with bug reports :)
>
> The Cygwin patches to 1.1.4 from Gerrit P. Haase have been merged in
> before 1.1.5 came out, but when I tried out 1.1.5 a few weeks ago, I had
> some problems with Cygwin's make utility. On the other hand, cygwin is a
> fast moving target, and chances are the problems are fixed now.
>
> If you are interested in making it happen, hop on the Kaffe mailing
> list, or the irc://irc.freenode.org:#kaffe channel.
>
> cheers,
> dalibor topic
>
Re: Windows support
Posted by Dalibor Topic <ro...@kaffe.org>.
Berlin Brown wrote:
> Hello, I am not a JVM guru like the others, just a long-time Java
> advocate and sit-back-and-watch Harmony person. I have a comment, I
> know there will be a strong interest in getting an Apache backed
> system on Linux/Solaris/Mac environments. It would be nice to see a
> strong effort for Windows support. One of the key selling points of
> Java is its portability. Having a Windows port would be key. And,
> we all love to hate Windows, but it is pretty pervasive. It would be
> nice if the Windows support was part of the initial focus and not an
> after thought like some of the research Open-Source/JavaVMs that are
> out now.
Hi Berlin,
I believe that mostly the reason why the current release of Kaffe has
very little Windows support (just via cygwin) was the lack of big enough
interest from developers familiar with and working on Windows. That, and
the lack of some volunteer who cared enough to drive the issue to
completion.
I'm glad that you are considering volunteering to fix that problem:
there are Kaffe ports to Cygwin, as well as a mingw32 port, and ports to
WindowsCE platforms as well as DOS. You do not have to be VM guru to
fix build problems, and there have even been patches floating around for
Kaffe to make it build fine with MSVC++. On a side note, integrating
MSVC++ into Kaffe's (and other autotools based) build systems should be
fairly easy using cccl m4 macros, I've done this myself in a different
project to have a single, automated build system.
On the other hand, for all I know IKVM works great on windows already. :)
> Also, I would suggest a port of C-Code Kaffe/GCJ JavaVM code to a
> Microsoft backed compiler, possibly Visual C++ 6/7. Or at least
> provide build support for the Windows platforrm.
Get cygwin, get kaffe 1.1.5 tarball, get jikes 1.22 tarball. unpack
jikes, configure && make && make install it, unpack kaffe, configure it
with --disable-native-awt (cygwin did not support gtk last time I
checked) and --disable-sound (neither did it support ALSA or ESD) and
--enable-pure-java-math (unless you installed GNU MP libs with headers).
make, and pester me with bug reports :)
The Cygwin patches to 1.1.4 from Gerrit P. Haase have been merged in
before 1.1.5 came out, but when I tried out 1.1.5 a few weeks ago, I had
some problems with Cygwin's make utility. On the other hand, cygwin is a
fast moving target, and chances are the problems are fixed now.
If you are interested in making it happen, hop on the Kaffe mailing
list, or the irc://irc.freenode.org:#kaffe channel.
cheers,
dalibor topic