Supporting thread for the SWTSwing (http://www.os2world.com/content/view/14799/71/) bounty.
Yesterday I've checked Eclipse 3.2.2 which works with Innotek Java 1.4.2 (doesn't work with Java 1.4.1 from GC).
There are some issues but it's quite usable already.
The following screenshots use the Lipstick Look'n'Feel http://regis.risp.pl/ (http://regis.risp.pl/) which is better looking than the default Swing Metal Look'n'Feel. The Eclipse package I created and sent to some testers has some more Look'n'Feels on-board (e.g. the jgoodies looks LAF http://www.jgoodies.com/freeware/looks/index.html (http://www.jgoodies.com/freeware/looks/index.html) and the NimROD LAF http://personales.ya.com/nimrod/index-en.html (http://personales.ya.com/nimrod/index-en.html)).
Here are again some Eclipse screenshots...
You can also find all images (including full-screen images) here: http://www.juergen-ulbts.de/Sonstiges/Java/Eclipse322SWTSwing/ (http://www.juergen-ulbts.de/Sonstiges/Java/Eclipse322SWTSwing/)
Create a new project:
(http://www.juergen-ulbts.de/Sonstiges/Java/Eclipse322SWTSwing/Eclipse322_SWTSwing_on_eCS20_02_small.jpg)
Adding a new class:
(http://www.juergen-ulbts.de/Sonstiges/Java/Eclipse322SWTSwing/Eclipse322_SWTSwing_on_eCS20_03_small.jpg)
Main class of the just created project:
(http://www.juergen-ulbts.de/Sonstiges/Java/Eclipse322SWTSwing/Eclipse322_SWTSwing_on_eCS20_04_small.jpg)
Please support the SWTSwing development and Java 1.4 backport!
Thank you!
Hi!
I'll release Azureus 2.5.0.4 (full package - startup script and app) and an Eclipse 3.2.2 (Runner package - Eclipse 3.2.2 Windows has to be downloaded from the official website) based on the latest SWTSwing shortly on my website.
Hmm to make sure; where do I download Azureus 2.5.0.4 from?
Hi
I hate to rain on anyone's parade but this statement
"Please support the SWTSwing development and Java 1.4 backport!"
seems a bit defeatist.
What is really required is a current java for eCS(OS/2) - too many java developers are now using java1.5 and moving to 1.6 making their apps unusable in an OS/2 based environment.
So, Sorry for the negative comments but it is *Not* backports of java software that are required but an up to date java environment.
Regards
Pete
I agree with Pete. We really, really need the latest version of Java. If this requires $$$ to get a developer to do this, I'm willing to part with $100 to start the ball rolling.
cytan
We actually have a bounty for the latest native java platform for OS2; the bounty can be found here (http://www.os2world.com/content/view/337/71/). It's currently value is 320 USD.
I know about the bounty which is really a hit and miss affair. To ensure that JAVA is ported, how much will it cost to hire a programmer to do this? Should this be something the community should seriously consider doing?
cytan
Well, I agree that it's a little bit tricky; if someone should take on to do the entire port them self they would either be really interested of the workload or a bounty would be really attractive. The current amount won't justify this enormous work and I would say that the current bounty sum needs to at least be 10 times to make it really interesting, ie 3000-4000 USD and there is currently a long way to go there.
To hire a programmer to do this would surely cost at the lowest 2-3 month of development time if everything needs to be done from scratch. To be said, I'm not a programmer, so I might be totally of on my guess. But, since earlier version has been released the question is how big of help that code base can be or at project knowledge. So the question is if there is any interest from either Golden Code or Innotek to release what ever material they can to the community to help it out....
Quote from: kimhav on 2007.08.23, 16:12:16
Hmm to make sure; where do I download Azureus 2.5.0.4 from?
I'll put it on my website once I've got some time to build the package.
I've just release the first public Eclipse Enabler package based on an unreleased Eclipse on Swing (EOS) plugin which uses the SWTSwing release from 2007-08-20.
Quote from: Pete on 2007.08.25, 16:34:50
Hi
I hate to rain on anyone's parade but this statement
"Please support the SWTSwing development and Java 1.4 backport!"
seems a bit defeatist.
What is really required is a current java for eCS(OS/2) - too many java developers are now using java1.5 and moving to 1.6 making their apps unusable in an OS/2 based environment.
So, Sorry for the negative comments but it is *Not* backports of java software that are required but an up to date java environment.
Regards
Pete
Pete, SWTSwing is only a backport not Eclipse 3.2.2. Using the term 'backport' with SWTSwing is not even correct. SWTSwing works better with Java 1.5 and even better with Java 1.6. SWTSwing has only been modified to fall back to older Swing methods that have been replaced or extended in Java 1.5 and 1.6 (e.g. Minimize to System tray function) if it is run on Java 1.4.
We can still use and still need SWTSwing even if we get Java 1.5, 1.6 or 1.7 on eCS and OS/2 because if we don't have SWT on your platform developers can't use Eclipse as Development IDE and it's currently the one most often used...not only for Java development (there are PHP,C,.. plugins available).
Quote from: kimhav on 2007.08.26, 19:33:15
The current amount won't justify this enormous work and I would say that the current bounty sum needs to at least be 10 times to make it really interesting, ie 3000-4000 USD and there is currently a long way to go there.
To hire a programmer to do this would surely cost at the lowest 2-3 month of development time if everything needs to be done from scratch. To be said, I'm not a programmer, so I might be totally of on my guess. But, since earlier version has been released the question is how big of help that code base can be or at project knowledge. So the question is if there is any interest from either Golden Code or Innotek to release what ever material they can to the community to help it out....
I would say you have to calculate 120000 USD for this task if you really would like to restart a native port.
Here is what I got back from Eric Faulhaber (Golden Code) last year (November 2006) when I asked Golden Code if they would help with a port...after Sun said they would open source Java.
Quote
------------------------------------------------------------------------
> I'm sure you've seen that Sun is starting to make OpenSource.
Yes, we are pleased to see Sun open source Java, as it is a good thing
for the platform, though we wish they had done it much sooner.
> if it's now
> possible to get an updated Java on OS/2 and eComStation.
> So far I think it's still unlikely as there are not enough developers.
Agreed. While the OS/2 community has (or at least had) many smart and
experienced developers, this simply is not a project that one can
"tinker" with in order to successfully make a port. It would be
surprising if a group of part-time developers could make a good result
in a reasonable time frame. The 1.4.1 effort took us over 20 person
years with a dedicated team of very senior developers working ridiculous
overtime hours.
> Could you think of contributing parts of the Java 1.4.1 sourcecode to
> an Java 6
> (or 7) OpenJava Netlabs.org project?
> I'm thinking about the JNI, Sound, Threading,... stuff and everything
> else that
> could help to get it of the ground.
We will *consider* any idea, but frankly, this is unlikely. The J2SE
port was unbelieveably costly for us. To date, the OS/2 community's
financial support of this project (measured by sales of our port) has
been nearly non-existent. This is not to say no one has bought it; we
definitely appreciate the support of customers like you. There are just
not enough of you for us to make a living in the OS/2 market.
So we certainly cannot spend any additional effort on another port
without some significant fees paid in advance. Using our current port,
the work itself would probably take 3-4 people from our team about 6
months (full-time) to finish, including testing (possibly more if a
shift from the ICC to GCC compiler was necessary). In the case of
GPLing our sources, such a fee would essentially be the last money we
would ever recover from our effort, so it would have to cover our costs
as a company for a half year or so. I just don't know where this money
would come from.
> It would be still possible for GoldenCode to offer paid support for
> companies
> that will use this OpenSource Java and there is also still the Sun Java
> certification program.
No, speaking from painful experience, it is simply not realistic for us
to make a viable business in the corporate market for OS/2 (software,
services...) anymore; it is truly dead. And even if it weren't dead,
there would be many people that would be potential competitors here once
the sources were GPL. The only way this would happen would be for us to
get paid up-front a substantial enough fee to justify the work. And it
would take significant work for us to GPL the sources, even if we didn't
take on the porting work ourselves, though knowing the sources as we do
I am positive that our team is the best fit for the porting work.
Sun has not yet released everything under GPL yet (they are scheduled to
be done by June 2007), so that limits things too. Note also that Sun
has not yet decided whether they will release their TCK under the GPL.
This is quite important to enable a compatible result. You would be
very surprised at the "hoops" that we had to jump through to pass all
30,000 tests.
> In two days Warpstock Europe is starting and the OpenSource Java will
> be also a
> topic there at the discussion rounds.
Sorry we don't have better news for you to report, but we must, first
and foremost, pay our people and pay our bills...
------------------------------------------------------------------------
Well if its $120k, then I think we can forget about ever having any JAVA port :-(.
Life sucks and then you die!
cytan
Thanks for the info about Golden Code; another question then; are there parts in the existing java that could be improved to make work better or last longer for our purpose or are we really stuck with what we got?
Quote from: kimhav on 2007.08.27, 12:09:49
Thanks for the info about Golden Code; another question then; are there parts in the existing java that could be improved to make work better or last longer for our purpose or are we really stuck with what we got?
As we don't have access to the code Innotek used to reuse the Windows Java and also no access to the Golden Code source code we are pretty much stuck right now.
I still hope that Golden Code will provide some pieces once someone is working on the Open Source Java from Sun. I know there is something happening with the Open Source Java and eCS right now (someone is taking a look) but it's to early to say more. Maybe on Warpstock this year there will be more info about this project.
I really think we need some parts from Golden Code to get a new Java Version build upon the Open Source Java to get a new version...if we want to have a full native port.
Quote from: magog on 2007.08.27, 12:22:50
I still hope that Golden Code will provide some pieces once someone is working on the Open Source Java from Sun. I know there is something happening with the Open Source Java and eCS right now (someone is taking a look) but it's to early to say more. Maybe on Warpstock this year there will be more info about this project.
It would be nice to get it put on Netlabs or somewhere, because there might be quite a few people looking at it. For example, I have JDK 6 source on my laptop. I've only looked at one file so far. It doesn't look impossible, but it is a very large task. It might be better to target JDK 7, as a port might be 2 years away.
I would tend to agree to look on the next release, if we're going wait, well let's wait for something really good. About Golden Code and that they don't feel like handing over the code, then again - what good does it do them that they have the code in a drawer? What can they loose on handing over code to for example netlabs? They said them self that they have gained almost nothing on the Java ports so what is there to loose then.
Quote from: kimhav on 2007.08.30, 10:06:55
I would tend to agree to look on the next release, if we're going wait, well let's wait for something really good. About Golden Code and that they don't feel like handing over the code, then again - what good does it do them that they have the code in a drawer? What can they loose on handing over code to for example netlabs? They said them self that they have gained almost nothing on the Java ports so what is there to loose then.
I would tend to agree to look on the next release, if we're going wait, well let's wait for something really good. About Golden Code and that they don't feel like handing over the code, then again - what good does it do them that they have the code in a drawer? What can they loose on handing over code to for example netlabs? They said them self that they have gained almost nothing on the Java ports so what is there to loose then.
[/quote]
It also doesn't do them any good to hand over the code. One reason they might not want to is that they might still be maintaining it for business customers, who might grab Netlabs changes and ask Golden Code to support them. I've seen this type of thing happen. Sometimes it has to be true abandonware before it can be released as a loss.
The $120,000 figure is also only about half the cost quoted by Golden Code to do the port. 3-4 developers for 6 months in the USA is about $240000. They probably had more than that in the initial work. I would have trouble giving that away, especially when they probably had much more OS/2 users saying they'd buy the product than actually did.
When was the last time you gave a half years salary to street beggars? :-[
Quote from: lwriemen on 2007.08.30, 15:36:47
...who might grab Netlabs changes and ask Golden Code to support them. I've seen this type of thing happen. Sometimes it has to be true abandonware before it can be released as a loss.
Red Hat, the Linux enterprise vendor, has its version of its Linux code snapped by Oracle (Unbreakable Linux) and by CentOS. Yet, Red Hat does no support those alternative distros --even if those are practically the same code. Yet Red Hat has remained unaffected financially and continues selling subscriptions supporting *only* its own distribution of Linux long after the analysts predicted its demise when Oracle, particularly, announced that it would rebrand Red Hat's Linux distro one year ago (2006, OracleWorld) into Oracle Unbeakable Linux.
Quote from: lwriemen on 2007.08.30, 15:36:47
The $120,000 figure is also only about half the cost quoted by Golden Code to do the port. 3-4 developers for 6 months in the USA is about $240000. They probably had more than that in the initial work. I would have trouble giving that away, especially when they probably had much more OS/2 users saying they'd buy the product than actually did.
When was the last time you gave a half years salary to street beggars? :-[
This kind of analogy reveals an usually unspoken hard-linked perspective in many OS/2ers From SSI to many of its users, and programmers (I have read Alex Taylor's posts elsewhere). Suffice it to say that the community of "beggars" have the potential to find bugs in Golden Code's excellent but old native Java implementation. Suffice it to say, also, that the community of "beggars" have the potential to add extensions to the code, improve the same and innovate it --at the same time adding back those improvements to the Golden Code native Java framework.
The latter could be cleansed and tested more extensebly by GCD and folded into their own stable OS/2 version offering that GCD would support by selling subscriptions and/or supporting their existing customers with updated software. GCD would not have to support the "raw" community based branch -- the Xen community based virtualization offering is not supported by XenSource nor VirtualIron (another vendor whose virtualization product is based on Xen).
The above is exactly what XenSource does with the open source Xen virtualization technology. We at Metztli Information Technology (http://metztli%20information%20technology) are a XenSource Certified Solution Provider --I know first hand of the process. As a matter of fact, variations of the above is how most open source companies base their business model on (you might want to search for the 2007 Open Source Business Conference (OSBC) in San Francisco for some of the slides presented by MySQL CEO on Business Models).
Further, in some open source business models proprietary code (taking care to honor the GPL or related open source licenses, as well as the obvious proprietary ones) is added to open source code (XenSource, EnterpriseDB, Novell's SuSE Linux Enterprise Desktop, etc.).
Possibly if the majority of OS/2ers' hard-linked perspective of viewing a community as "beggars" is overcome, OS/2 will rebound to become an active competitor rather than an passive one. If that proprietary notion is ever overcome, organizations like SSI and GCD would rediscover that what is given to the community would be returned in a at least an equal amount --if not more-- provided care is taken on how to tap into that community's activity; the latter would enable those organizations with the possibility of creating an current business model *around* open source (because) open source in itself *is not* a business model.
Quote from: El Vato on 2007.08.31, 02:25:20
This kind of analogy reveals an usually unspoken hard-linked perspective in many OS/2ers From SSI to many of its users, and programmers (I have read Alex Taylor's posts elsewhere). Suffice it to say that the community of "beggars" have the potential to find bugs in Golden Code's excellent but old native Java implementation. Suffice it to say, also, that the community of "beggars" have the potential to add extensions to the code, improve the same and innovate it --at the same time adding back those improvements to the Golden Code native Java framework.
The latter could be cleansed and tested more extensebly by GCD and folded into their own stable OS/2 version offering that GCD would support by selling subscriptions and/or supporting their existing customers with updated software. GCD would not have to support the "raw" community based branch
Well said. You are of course right; open source doesn't have to mean free (beer). The dynamics of open source are different on OS/2. The user base and developer pool are smaller, and the operating system isn't open. The latter is really too bad, because OS/2 could really kick Linux's butt in the embedded market.
Quote from: lwriemen on 2007.08.31, 04:25:59
Well said. You are of course right; open source doesn't have to mean free (beer). The dynamics of open source are different on OS/2. The user base and developer pool are smaller, and ...
...precisely. There the reason that whatever can be made open, in this case GCD's OS/2 Java source, would be an positive influx of energy to the OS/2 cause, as Kim H. argued priorly.
Notice that I did not mention Innotek's Odin-based Java hack –that is an step back in the true OS/2 evolution. That hack effectively affected GCD's true potential for its continued survival by Innotek offering a "free" non-enterprise grade layered approach whose quality never matched that of GCD's.
Yes, the initiator of this post might make Eclipse walk three steps forward with Innotek's OS/2 pseudo port of Java, but that is it. Even if the whole Eclipse IDE were to function, its true functionality is dependent on the plugins available for the IDE and Innotek's Java simply is not stable enough for that. Eclipse without its plugins for the Tomcat Server/WebSphere Community Edition and other Java applications is the equivalent of writing code in a Java editor. That aspect of the plugins is not a peachy issue to resolve, as can be deduced from attempting to start the Tomcat Server --from within the Sun Microsystems' Netbeans IDE 4.x and later-- to test an application that a programmer is currently developing.
On the other hand, I appreciate your civilized response to my argument. Refreshing, indeed. In the past, when I advanced a certain notion to the open source of this or that effect, it was called "stupid" by so called "expert OS/2 consultants," and pseudo programmers; at the same time those defended to death the proprietary shadows portrayed on their cave walls by the preachers of pre Internet business models –inadequate by todays technologically changed landscape where the OS/2 finds itself.
As a matter of fact, I also advanced the notion on which this thread is based (SWTSwing) in a certain Java thread that effectively was ignored... Might the OS/2 user base be undergoing a metamorphosis and preparing for the operating system cross pollination? "Could be, Capitan!" the Q might have answered, as a Mariachi band played behind him :D.
Best-
Not to take away from the thread, but wow, you write well :o
I have to disagree on your statements about innotek's java "patch"
I have always found that their java hack not only ran more java apps but was also much more stable then GCD's. The real problem with GCD wasn't the fact that they wanted money for it, but it didn't live up to the hype which they created around it. At the very least innotek's approach gave us a tested and supported vm (windows) as much as I was originally against it, it worked. The problem was more that it wasn't a sustainable approach, however if GCD had taken the time the innotek bought them and used it to further develop their VM and bring it to a level above innotek then GCD would be a viable option today. However, the biggest fault with the whole thing was that GCD expected to be able to finance development of the product (I am not assuming they wanted to turn a profit on it), and simply put you will not be able to break even on any large codebase OS/2 project unless it was platform neutral to being with (and it still would be hard).
GCD being open source would only help us if: Their patches could be applied to the source for a newer vm, assuming you could get the sun source for a newer vm, and if we could find sufficient developers to make headway on it, now this isn't a matter of someone being able to "port" they would need to be able to integrate platform specific features into the vm, and have the time to do it.
We should not look down on what innotek has done for us, just because you believe their idea is flawed. There are plenty of native linux apps out there today which have wine sitting in their code somewhere. It would have been nice if odin and the innotek wrapper could have been made to integrate more closely with OS/2.
I also have to argue there is no way to step back as far as OS/2 is concerned. If something is created it can only be a step forward. There is no path of development, the OS is basically frozen circa 1994 and I will gladded accept any application which is provided, whether that be a native application, java, emulated, ported etc..
Quote from: El Vato on 2007.08.31, 08:08:45
Quote from: lwriemen on 2007.08.31, 04:25:59
Well said. You are of course right; open source doesn't have to mean free (beer). The dynamics of open source are different on OS/2. The user base and developer pool are smaller, and ...
...precisely. There the reason that whatever can be made open, in this case GCD's OS/2 Java source, would be an positive influx of energy to the OS/2 cause, as Kim H. argued priorly.
Notice that I did not mention Innotek's Odin-based Java hack –that is an step back in the true OS/2 evolution. That hack effectively affected GCD's true potential for its continued survival by Innotek offering a "free" non-enterprise grade layered approach whose quality never matched that of GCD's.
Yes, the initiator of this post might make Eclipse walk three steps forward with Innotek's OS/2 pseudo port of Java, but that is it. Even if the whole Eclipse IDE were to function, its true functionality is dependent on the plugins available for the IDE and Innotek's Java simply is not stable enough for that. Eclipse without its plugins for the Tomcat Server/WebSphere Community Edition and other Java applications is the equivalent of writing code in a Java editor. That aspect of the plugins is not a peachy issue to resolve, as can be deduced from attempting to start the Tomcat Server --from within the Sun Microsystems' Netbeans IDE 4.x and later-- to test an application that a programmer is currently developing.
On the other hand, I appreciate your civilized response to my argument. Refreshing, indeed. In the past, when I advanced a certain notion to the open source of this or that effect, it was called "stupid" by so called "expert OS/2 consultants," and pseudo programmers; at the same time those defended to death the proprietary shadows portrayed on their cave walls by the preachers of pre Internet business models –inadequate by todays technologically changed landscape where the OS/2 finds itself.
As a matter of fact, I also advanced the notion on which this thread is based (SWTSwing) in a certain Java thread that effectively was ignored... Might the OS/2 user base be undergoing a metamorphosis and preparing for the operating system cross pollination? "Could be, Capitan!" the Q might have answered, as a Mariachi band played behind him :D.
Best-
Quote from: Robert Deed on 2007.08.31, 17:46:15
I have to disagree on your statements about innotek's java "patch"
I have always found that their java hack not only ran more java apps but was also much more stable then GCD's [...]
We could argue in an abstract manner (without quantifying by actual test results on equal hardware and OS deployments) but that would not get us anywhere.
But let us analyze each perspective:
You: Desktop user applications (relatively frivolous by default)
Myself: Enterprise grade applications (stability, performance, and SMP capable)
Quote from: Robert Deed on 2007.08.31, 17:46:15
The real problem with GCD wasn't the fact that they wanted money for it, but it didn't live up to the hype which they created around it. At the very least innotek's approach gave us a tested and supported vm (windows)[...]
An desktop user will resist buying an product if an alternative (even though of lesser quality) is available. Remember the free offering of IE that killed Netscape as a business entity? Obviously, the idea was recycled.
Quote from: Robert Deed on 2007.08.31, 17:46:15
GCD being open source would only help us if: [...]
This is the proprietary perspective that still haunts many OS/2ers! If the open source innovators/developers placed conditions on the community before releasing their software, open source would not be the phenomenon it is today.
That presumption that there has to be an effect(waiting developers) a priori justifying for the developer/innovator to release his/her cause(code) is endemic of proprietary business models –intrinsically deterministic and against the principles of free software --the necessary precursor of open source. Let go of it if you really want OS/2 to bubble up.
There is an old but known Greek argument by Aristotle that is at the foundation of democratic thought. Paraphrasing: an elite class (developers for the purposes of this analogy) will not make the outcome of a society(OS/2 activity) the best that it can be; the inclussion of the other classes(those non specialists, i.e, users) in the society are required to make the outcome of that mingling of human elements the best society (project) it can be. (Obviously, the so called western democracies are not practicing those principles).
By projecting your perception on conditions that must preexist before something occurs you are limiting the task. The community= (developers + users) multiplicity of perspectives is greater than the individual one.
Quote from: Robert Deed on 2007.08.31, 17:46:15
We should not look down on what innotek has done for us, just because you believe their idea is flawed. There are plenty of native linux apps out there today which have wine sitting in their code somewhere. It would have been nice if odin and the innotek wrapper could have been made to integrate more closely with OS/2.
Yes. Wine is there for desktop users (frivolous apps)... I have successfully run a Flash 8 based screen saver on my Warp server < http://metztli-it.com/blog/2007/08/lotus_connections_screen_saver.html >.
Quote from: Robert Deed on 2007.08.31, 17:46:15
I also have to argue there is no way to step back as far as OS/2 is concerned. If something is created it can only be a step forward. There is no path of development, the OS is basically frozen circa 1994 and I will gladded accept any application which is provided, whether that be a native application, java, emulated, ported etc..
Any "solution" that inserts an additional layer of abstraction is a step backward.
It amounts to what MS strategy is by their attempt to demote Linux as a guest in the datacenter, providing as the "stable" base their WinXX server; it amounts to stopping further development of OS/2 as a host for the VPC application, an expendable solution that will not force either of the implementators (Innotek and/or MS) to commit any serious resources (as GCD did).
The native interface of the OS/2 is being "tricked" by an intermediary that will cause a performance hit to the OS/2. Again, it is a quick fix for the desktop user meanwhile the implementators find another gig . A disaster for us who regard (and expect) OS/2 as an enterprise grade beast to run demanding applications.
I just hope that you will not reply simply for the sake of argument that, "Why would you want to use OS/2 for development?" as I recall someone once replied to me when I advanced the former notions (sigh).
Quote from: Robert Deed on 2007.08.31, 17:46:15
We should not look down on what innotek has done for us, just because you believe their idea is flawed. There are plenty of native linux apps out there today which have wine sitting in their code somewhere. It would have been nice if odin and the innotek wrapper could have been made to integrate more closely with OS/2.
In fact, originally GCD and Innotek were going to work on this together and use the wrapper approach. If GCD had not backed out of that deal then we would have had their efforts put into the wrapper so that it could be used for even more things. I listened to GCD's presentation at WS 2002 and they said it would have been as much work to get Odin up to par as to do the native port. The difference though, once Java was ported we had Java (and not as up-to-date a version at the time as the wrapper version) whereas, with the wrapper approach we also got adobe, flash, and OpenOffice. If the work that was directed by GCD into the port had been put into the wrapper there would be better integration of the wrapper into OS/2. It is their business, and they can put their resources where they desire but the statement by El Vato "That hack effectively affected GCD's true potential for its continued survival by Innotek offering a "free" non-enterprise grade layered approach whose quality never matched that of GCD's." forgets that GCD and Innotek were originally partners and GCD decided to go its own way and despite its claim that it could keep up with releases better with the "Native" approach, its first release was not even up to the current Innotek release and they never made the attempt.
If their release had been at least 1.4.2 they would have had a better chance to compete but their inability to do so (whether it was financial or technical) suggests that their approach was the flawed one. If they had went with the wrapper approach, they might have been able to bring in other applications as well. Unfortunately, the changes that were made to Odin (Innotek Wrapper) are not open for us to make use of them.
Andy
Quote from: El Vato on 2007.09.01, 02:40:29
Any "solution" that inserts an additional layer of abstraction is a step backward.
This is an absolute truism. Unfortunately, too many software developers these days think that abstraction layers are a requirement for reuse. No matter how thin a layer is, it introduces a performance hit and some added complexity. The real path is to program at a higher level of abstraction.
I actually don't know where you got that I was a desktop user. Unfortunately I am realistic enough to know that when I have a need for "enterprise level java" support, I don't use OS/2. Especially when no one has ever brought java enterprise edition to OS/2 in a complete form, that includes GC..
Actually, as an enterprise user I would refuse to purchase a third party contribution which passed almost all sun's tests with a few contradictions and was outdated the day it was released.
There are things which have to exist for something to happen. The world is driven by supply and demand, unfortunately as far as OS/2 is concerned there will never be enough demand to warrant the expense to supply the product. It is a vicious circle which has plagued the community many times over. It also is that every time an ISV comes out with someone they feel the need to bash the only other options. This may work in Windows where there is a (figuratively) infinite supply of users, however in the OS/2 community it hurts the overall market. GCD bashed the Innotek approach repeatedly but when asked to show some hard facts there was a vacuum.
Anyway. If someone ports SWTSwing, and the end results is an application which currently will not work, will atleast load it is a step in the right direction. Nothing we do is meant to be the end all and be all of everything. it is just making things slightly better.
Hell.. personally.. I have given up on OS/2 as a development platform (unless I am writting an OS/2 application)/
Quote from: El Vato on 2007.09.01, 02:40:29
Quote from: Robert Deed on 2007.08.31, 17:46:15
I have to disagree on your statements about innotek's java "patch"
I have always found that their java hack not only ran more java apps but was also much more stable then GCD's [...]
We could argue in an abstract manner (without quantifying by actual test results on equal hardware and OS deployments) but that would not get us anywhere.
But let us analyze each perspective:
You: Desktop user applications (relatively frivolous by default)
Myself: Enterprise grade applications (stability, performance, and SMP capable)
Quote from: Robert Deed on 2007.08.31, 17:46:15
The real problem with GCD wasn't the fact that they wanted money for it, but it didn't live up to the hype which they created around it. At the very least innotek's approach gave us a tested and supported vm (windows)[...]
An desktop user will resist buying an product if an alternative (even though of lesser quality) is available. Remember the free offering of IE that killed Netscape as a business entity? Obviously, the idea was recycled.
Quote from: Robert Deed on 2007.08.31, 17:46:15
GCD being open source would only help us if: [...]
This is the proprietary perspective that still haunts many OS/2ers! If the open source innovators/developers placed conditions on the community before releasing their software, open source would not be the phenomenon it is today.
That presumption that there has to be an effect(waiting developers) a priori justifying for the developer/innovator to release his/her cause(code) is endemic of proprietary business models –intrinsically deterministic and against the principles of free software --the necessary precursor of open source. Let go of it if you really want OS/2 to bubble up.
There is an old but known Greek argument by Aristotle that is at the foundation of democratic thought. Paraphrasing: an elite class (developers for the purposes of this analogy) will not make the outcome of a society(OS/2 activity) the best that it can be; the inclussion of the other classes(those non specialists, i.e, users) in the society are required to make the outcome of that mingling of human elements the best society (project) it can be. (Obviously, the so called western democracies are not practicing those principles).
By projecting your perception on conditions that must preexist before something occurs you are limiting the task. The community= (developers + users) multiplicity of perspectives is greater than the individual one.
Quote from: Robert Deed on 2007.08.31, 17:46:15
We should not look down on what innotek has done for us, just because you believe their idea is flawed. There are plenty of native linux apps out there today which have wine sitting in their code somewhere. It would have been nice if odin and the innotek wrapper could have been made to integrate more closely with OS/2.
Yes. Wine is there for desktop users (frivolous apps)... I have successfully run a Flash 8 based screen saver on my Warp server < http://metztli-it.com/blog/2007/08/lotus_connections_screen_saver.html >.
Quote from: Robert Deed on 2007.08.31, 17:46:15
I also have to argue there is no way to step back as far as OS/2 is concerned. If something is created it can only be a step forward. There is no path of development, the OS is basically frozen circa 1994 and I will gladded accept any application which is provided, whether that be a native application, java, emulated, ported etc..
Any "solution" that inserts an additional layer of abstraction is a step backward.
It amounts to what MS strategy is by their attempt to demote Linux as a guest in the datacenter, providing as the "stable" base their WinXX server; it amounts to stopping further development of OS/2 as a host for the VPC application, an expendable solution that will not force either of the implementators (Innotek and/or MS) to commit any serious resources (as GCD did).
The native interface of the OS/2 is being "tricked" by an intermediary that will cause a performance hit to the OS/2. Again, it is a quick fix for the desktop user meanwhile the implementators find another gig . A disaster for us who regard (and expect) OS/2 as an enterprise grade beast to run demanding applications.
I just hope that you will not reply simply for the sake of argument that, "Why would you want to use OS/2 for development?" as I recall someone once replied to me when I advanced the former notions (sigh).
That would be true if odin was a 100% abstraction. While some of it is an asbraction, some of it is filling in missing api's or replacing existing api's with different function.
If this is your argument, java itself is a step backwards and we are fortunante there isn't a newer release.
And as far as performance is concerned, in most java benchmarks the innotek abstracted step backwards approach was actually the superior performer. You can say things like the sun supplied vm was not supplied to gc or something to that extent (which GC used to argue).
Quote from: lwriemen on 2007.09.01, 04:16:34
Quote from: El Vato on 2007.09.01, 02:40:29
Any "solution" that inserts an additional layer of abstraction is a step backward.
This is an absolute truism. Unfortunately, too many software developers these days think that abstraction layers are a requirement for reuse. No matter how thin a layer is, it introduces a performance hit and some added complexity. The real path is to program at a higher level of abstraction.
Quote from: Robert Deed on 2007.09.01, 04:28:43
That would be true if odin was a 100% abstraction. While some of it is an asbraction, some of it is filling in missing api's or replacing existing api's with different function.
If this is your argument, java itself is a step backwards and we are fortunante there isn't a newer release.
First off, my comment was a general statement and not directed at Innotek vs. Golden Code Java. You are right; Odin isn't an abstraction layer. Odin is pretty much native code.
Java would have been just another programming language if it hadn't been for Microsoft's predatory monopoly. It may have garnered the same amount of success, but it will always be slower than a native application. IBM tried to alleviate this a little with SWT.
Poor software development practices are also the reason open source became popular. The millions of eyes argument would be rather specious if high quality code were the norm rather than the exception. :o
Quote from: Robert Deed on 2007.09.01, 04:24:36
I actually don't know where you got that I was a desktop user.
Look inside you and your questions should be answered...
Quote from: Robert Deed on 2007.09.01, 04:24:36
Unfortunately I am realistic enough to know that when I have a need for "enterprise level java" support, I don't use OS/2. Especially when no one has ever brought java enterprise edition to OS/2 in a complete form, that includes GC..
You might try the GCD on your SMP machines –after you find out that Innotek's Odin based implementation locks those (machines), if it starts at all.
Quote from: Robert Deed on 2007.09.01, 04:24:36
Actually, as an enterprise user I would refuse to purchase a third party contribution which passed almost all sun's tests with a few contradictions and was outdated the day it was released.
This simply shows your lack of understanding of the subject that I am arguing. In organizations whose resources are tied to a large vendor's implementation of Java, those Java deployments lag behind current development. Why? Because what is required is stability and performance in those commercial grade deployments. That newest Java version has to have the bugs ironed out before it is acceptable to replace the year or two older Java framework that continues supporting that business infrastructure.
On the other hand you are not alone in your misperception that the most current and with most whistles Java implementation should be the rule. I have met a few IT people in corporations who still, "do not understand why their WebSphere deployments are still running under version 1.4.12."
Quote from: Robert Deed on 2007.09.01, 04:24:36
There are things which have to exist for something to happen.
Again, reread what I wrote before and elaborate carefully. Likely you will reach the (painful ?) conclusion that what you have been shown on the wall are shadows of a bygone pre Internet era where there where no notions of communities because of the silos effectively blocking any potential collaboration.
Quote from: Robert Deed on 2007.09.01, 04:24:36
The world is driven by supply and demand, unfortunately as far as OS/2 is concerned there will never be enough demand to warrant the expense to supply the product. It is a vicious circle which has plagued the community many times over.
There has never been an OS/2 community in the sense of those that exist in the open source arena. The closest it has come is to a client-server relationship (i.e., vendor-user). In this client-server scenario the collaborative and interactivity of an open source (community) equivalent is non existent. In its place there is an inflexible and mostly one way relationship from the server(vendor) to the client(user).
Quote from: Robert Deed on 2007.09.01, 04:24:36
It also is that every time an ISV comes out with someone they feel the need to bash the only other options. This may work in Windows where there is a (figuratively) infinite supply of users, however in the OS/2 community it hurts the overall market. GCD bashed the Innotek approach repeatedly but when asked to show some hard facts there was a vacuum.
Again, start your Java application in your SMP machine –surely, there will be a (digital) vacuum left by the application failure.
Quote from: Robert Deed on 2007.09.01, 04:24:36
Anyway. If someone ports SWTSwing, and the end results is an application which currently will not work, will atleast load it is a step in the right direction. Nothing we do is meant to be the end all and be all of everything. it is just making things slightly better.
I thought that your pinched nerve was due to my criticism on Innotek's inadequate Java hack –What say you?!
Quote from: Robert Deed on 2007.09.01, 04:24:36
Hell.. personally.. I have given up on OS/2 as a development platform (unless I am writting an OS/2 application)/
Indeed, you are an desktop user by your own admission...
Quote from: abwillis on 2007.09.01, 03:27:20
[...] I listened to GCD's presentation at WS 2002 and they said it would have been as much work to get Odin up to par as to do the native port. [...] but the statement by El Vato "That hack effectively affected GCD's true potential for its continued survival by Innotek offering a "free" non-enterprise grade layered approach whose quality never matched that of GCD's." forgets that GCD and Innotek were originally partners and GCD decided to go its own way and despite its claim that it could keep up with releases better with the "Native" approach, its first release was not even up to the current Innotek release and they never made the attempt.
I was not at that public relations presentation, Andy. Accordingly I can not engage in the argumentative process of "he said that...but she that. Or they said that."
Even if I had been there, quite possibly my take on the issue would likely differ from yours. We are all preconditioned to grasp only what we want to hear and filter out other facts (that is how politicians grasp power to manipulate populations and it happens in the most vociferous self described democratic societies).
I am commenting on what is explicit now by analyzing the outcome of forces that shaped the current situation in which one important resource of the OS/2, as is Java, has been mangled.
Eclipse is the basis for such enterprise grade applications such as IBM's Rational software suite (IDE, specifically); it is also the basis of BEA's competitive product offering, and others.
Accordingly, the argument is that the ends(enterprise grade app) should justify the means (Java). My position is that Innotek Java OS/2 implementation simply does not offer the required stability due to the Odin layer approach.
Evidently, I am using the former statement as an supportive clause for GCD to consider open sourcing their excellent Java implementation as --Kim H. hinted before.
On the other hand, you and others are equally free to request Innotek to free their Java hack --where is the problem?.
"As a man's life approaches the summit, the cold increases [affinity to ephemeral creations] also increase, evidently."
Within the next couple of days I'll release an updated Eclipse Enabler Version (should be v1.1) with the latest SWTSwing + EOS Plugin.
UPDATE:
I just released the Eclipse Enabler v1.1.0 (2007-10-03) on my website.
Eclipse is starting much faster now!!
http://www.juergen-ulbts.de/swtswing (http://www.juergen-ulbts.de/swtswing)
I also hope to get the special Azureus build out today...
Yesterday evening I created the first public Test version of Azureus4eCS.
You can get Azureus4eCS v0.1.0 on my website. It includes everything needed to run Azureus v2.5.0.4 on eCS and OS/2.
I hope to get some reports.
In the package is a configuration instruction. Please follow it carefully.
http://www.juergen-ulbts.de/swtswing (http://www.juergen-ulbts.de/swtswing)
Hello magog,
I have just installed your azureus4eCS package and so far I can tell it works OK.
Is there anything you like me to test?
I have installed intermediate version, else I have followed your guide.
I guess I was a little to quick.
I have observed following things:
- when I go with mouse pointer over the face in front of a torrent listed, the entire screen goes gray, when I click somewhere on "desktop" all is normal again
- When I try to add a file the directories/files list is not working with all the directories, can change but do not see the contens of the directory
Thanks for all the work.
Regards
Smiley
Quote from: smiley on 2007.10.04, 16:35:40
Hello magog,
I have just installed your azureus4eCS package and so far I can tell it works OK.
Is there anything you like me to test?
Maybe you could look if you get problems when you have two or more parallel high speed downloads.
Quote from: smiley on 2007.10.04, 16:35:40
I have installed intermediate version, else I have followed your guide.
I guess I was a little to quick.
I have observed following things:
- when I go with mouse pointer over the face in front of a torrent listed, the entire screen goes gray, when I click somewhere on "desktop" all is normal again
- When I try to add a file the directories/files list is not working with all the directories, can change but do not see the contens of the directory
I assume you used Innotek Java. Is that correct?
-You might want to try another Look'n'Feel with Azureus. Currently the Lipstick Look'n'Feel is activated, maybe that's the problem.
In order to select another one you need to modify the startos2.cmd and/or startos2debug.cmd.
At the bottom I listed some Look'n'Feels that are available.
If you have problems with it contact me by E-Mail (go to my website http://www.juergen-ulbts.de/ (http://www.juergen-ulbts.de/) where you find a 'Contact' page).
[Edit] Seems to be an SWTSwing bug. It's calculating the wrong size for the popup text panel.
-I need to check the directory listing. Are directories without spaces working and directories with spaces not, or don't you see some similarity?
I also need to know which dialog you mean (there are several). Is it the Open Torrents Dialog?
I've just seen that is also doesn't refresh the content of the directory. You can get around this is you enter "*.*" in the entry field "File Name" and press enter. Then you should see all files.
I hope that helps.
Quote from: smiley on 2007.10.04, 16:35:40
Thanks for all the work.
Regards
Smiley
No problem, thank you for testing.
Nice to hear it's working beside some minor problems.
Hello magog,
I have been further investigating...
The thing with whole screen gray happens with all look&feels, just the color is different.
As for directory listings, there is no rule aparent, some drives will not list some directories with 3 letters, but directory with 8 will etc.
If I enter "*" in the filename box all is working fine. Its with open torrent, add file.
Another thing I get almost no download, with the same torrent that is downloading in win OK, the port seems to be OK.
Regards
Smiley
Quote from: smiley on 2007.10.05, 11:43:02
Hello magog,
I have been further investigating...
The thing with whole screen gray happens with all look&feels, just the color is different.
As for directory listings, there is no rule aparent, some drives will not list some directories with 3 letters, but directory with 8 will etc.
If I enter "*" in the filename box all is working fine. Its with open torrent, add file.
Another thing I get almost no download, with the same torrent that is downloading in win OK, the port seems to be OK.
Regards
Smiley
The full screen popup is indeed a bug in SWTSwing and I've informed Christopher Deckers via IM and also added it as bug at SourceForge (https://sourceforge.net/tracker/index.php?func=detail&aid=1807919&group_id=83103&atid=568338 (https://sourceforge.net/tracker/index.php?func=detail&aid=1807919&group_id=83103&atid=568338))
Do you use a hardware router or something like ISDNPM on the same computer. I also need some hardware specs from you.
So far I've not seen problems with the max speed but after some time the Azureus generates errors and I think it's Java NIO related. That's why I called it "test version". ;D
I need to look into this but it's hard to debug and I'm currently out of free-time.
By the way a german tester also reported slow download speed but I don't have details yet. All I can say is that it works pretty fast here (Athlon64 X2 3800+ with 1 GB RAM).
I have an IBM Thinkpad Z60m 1GB RAM 2GHz processor intel Pentium,
I have a hardware router connected to ADSL 1024/256, with UPnP enabled and no other machine on the router.
Broadcom integrated NIC with Broadcom driver 9.07 I believe.
I know this is the test version, so I am testing it. :)
And I hope the results of the tests could be useful.
In total I've got 209 downloads in 5 days of the Azureus4eCS package with minimal feedback (3 people).
- One SWTSwing bug found (popup as full screen window)
- Problem with slow download speed reported (not sure if this is an error or not - need more feedback)
- Problem with File Open dialog reported and confirmed.
- Golden Code Java 1.4.1 seems to run Azureus a little bit smoother (reported by one user)
@smiley
Thank you for your system specs
An update to my testing.
The download is working fine, guess it was just some "slow" torrents, or problematical trackers.
Quote from: magog on 2007.10.09, 12:35:58
In total I've got 209 downloads in 5 days of the Azureus4eCS package with minimal feedback (3 people).
3 people provided feedback? Wow that's much more than I generally receive!
209 downloads is also pretty good - most eCS/OS/2 users don't seem to like giving feedback - I'd like to think no news == good news but it's still disconcerting....
I must add Azereus to my list, however Dink's tor.exe + Python + bittorrent works pretty nicely here.
Cheers,
Paul.
Quote from: Paul Smedley on 2007.10.18, 05:47:41
Quote from: magog on 2007.10.09, 12:35:58
In total I've got 209 downloads in 5 days of the Azureus4eCS package with minimal feedback (3 people).
3 people provided feedback? Wow that's much more than I generally receive!
209 downloads is also pretty good - most eCS/OS/2 users don't seem to like giving feedback - I'd like to think no news == good news but it's still disconcerting....
I must add Azereus to my list, however Dink's tor.exe + Python + bittorrent works pretty nicely here.
Cheers,
Paul.
Hi Paul!
Well it I think I now received feedback from 6 people.
The download hit the top quite fast after the release.
I'm not at 253 downloads for the Azureus v0.10 package and 160 downloads for the Eclipse Enabler v1.1.0 package!
Just try Azureus, it won't blow up your machine. ;)
@ALL
It also doesn't hurt to click on the banner when you find something. I'm not getting rich but it could help to pay my website for at least 4 month out of 12 (not including www.jmdb.de).
For this month 10 clicks generated 2,14 USD (now you know what I mean by not getting rich). ;)