Open-Sourcing Java: because...

First of all I would like to thank everybody that posted some comments to my previous questions about open-sourcing Java. Now that I have different points of view I can comment on them. Of course these are my comments, not Sun's ;-) There we go:

Fast, fast, fast

Some comments state that Sun is slow solving issues ("you have to wait Sun to take your submission", "Sun react too slow to big issues", etc.)

Well, I don't know what slow is. I mean, that's quite a subjective question. Lots of people think Debian is slow, for instance. I disagree: Debian is released whenever Debian is ready. Nor sooner nor later. Of course if you want to deliver good quality software you need some time. I don't mean that Good Software Takes Ten Years. Get Used To it, but you need some time.

Mono, for instance, has taken some years to build 1.0. And it seems they're not up yet with Microsoft .NOT stuff (well, that's a moving target, so it's difficult).

Lots of open-source people get used to the "release-often" approach. I don't mean it's a bad method, it's just a way to do things. That's ok whenever your users are beta testers: you can detect bugs easily. Furthermore, most users get used to have to keep track manually of dependencies, I mean: lots of open-source projects depend on other open-source projects which are in turn released often, so whenever you want to upgrade one you have to upgrade different others, and this is a nightmare of upgrades.

I wouldn't like that to happen with Java. It's a pain in the neck. I just prefer to have a new Java version whenever the version is ready. And if it has any bugs in it I'd like them to be publicly available, say in a web page or something, so I can go and see if there's a problem there. And I'd like them to be solved in the next Java release. In such a big thing as Java is (with lots of libraries there) is difficult to solve all bugs. There's no software without bugs. All I want is to be able to vote for bugs so the more important to me are solved. So people can concentrate in those bugs annoying me.

And that's exactly what's happening with Java. So I don't see a reason to change it.

JCP is open to open source

Some other comments state that you have to belong to a Big Company to participate in the Java Community Process. I know that the Apache Foundation and JBoss, Inc. are big companies, but they are part of the Open Source Movement, so I assume the Open Source Movement is participating in the JCP.

So JCP is open to open source.

But it's open as well to individuals. To experts.

All that is required is to do some work at the JCP. I mean, if you become an expert I assume you are supposed to do some real work, not just be there hanging around at your spare time. (And this is just my opinion: I'm not an expert at the moment, mainly because I cannot do any more real work at all ;-) ).

Saving Sun's money

Some of the comments state that open sourcing Java is a money-saving technique for Sun ("this will cost Sun a lot").

Now I understand the reason why IBM is pushing to open-source Java !! ( ;-) )

Now, seriously, I think Sun people is smart enough to decide on how to save money by themselves. No help from IBM needed at the moment. They could start by reducing the funds for R&D to a minimum, to the same level as other companies. But then we wouldn't have HotSpot, those nice featured, high-technology garbage collectors, not even a Real Time Java, nor a sound, secure, operating system.

And that wouldn't be good for customers, nor for Java users, so it's a Bad Idea: I still want Sun to deliver free hardware and free StarOffice licenses to the Open Source and Academic Communities. I still want Sun to invest huge amounts of money to build the Best Of The World garbage collection technology I can use in my laptop. Making Sun save money is, indeed, a Bad Idea.

Open-Sourcing Java will reduce the bugs in J2EE

"They did a lot of work on the J2EE standard, and EJB's ended up being a big pain in everyone's butt." Well, J2EE and EJBs have nothing to do with open-sourcing Java. Both J2EE and EJBs are being worked on at the Java Community Process, which is (as far as I know) already open to everybody.

Distributing the JDK with Linux

Someone mentions that the JDK cannot be installed by default in a Linux distribution. Well, this is a major problem to me too. I would like it to be installed everywhere. For free. With no licensing restrictions. I also think Sun should fix that.

Kaffe and GnuClasspath

I have to agree that Sun should make things easiers for these groups to verify their implementations. In fact it would be great if someone at Sun could collaborate with them in their implementations. That's something Sun should consider, I think.

Conclusion

I still cannot see a real reason why Java should be open-sourced. In fact I think there's no real reason to open-source Java. Even reading Eric S. Raymond's Open Letter I can see no real reason at all.

"But Sun has done other things that make us wonder if the vision and courage to choose the open-source path are really there.", Eric says. Well, I do think Sun has vision and courage to go through the open-source path, mainly because it is already in the open-source path. And this vision has been Sun's vision for quite a long time now. We're not new to Open Source. We've open-sourced lots of things. We're open-sourcing Solaris. We're open-sourcing Java3D. We're open-sourcing Looking Glass. We're not new to Open Source.

" if you're serious about preparing Sun for the future we can all see coming in which code secrecy and proprietary lock-in will no longer be viable strategies ", Eric says. Lock-in you say? Does JDBC lock you in in any proprietary database? Does J2EE lock you in any proprietary application server? Does Java lock you in any proprietary operating system? There's a misunderstanding here somewhere. We're about open systems and open source, not the other way round. And we have always been, and, for sure, we will ever be.

Jean-Christophe Collet's blog expresses it quite well: Sun is all about open-source and open-systems, Sun is an open-source ally. It's \*the\* open-source ally. Sun will always be an open-source friend.

And that's one of the reasons why I like this company!!

Comentarios:

It's .NET :-)

Enviado por Cyrus Najmabadi en julio 09, 2004 a las 05:20 PM CEST #

"I still cannot see a real reason why Java should be open-sourced. In fact I think there's no real reason to open-source Java." And I see less and less reason to use Java. Mono is grabbing more and more of the high level language side of Linux. I'm not playing the troll here either -- as you might think from me mentioning Mono. I'm just rather mystified that Sun people continue to think that Open Source = wild and uncontrolled development and forking = terrible thing. Mono is developed mostly by Novell, and yet there's no "sign up for membership of the Mono development process", or requests to see your qualifications before you can submit code for consideration. No membership fees or bureaucracy surrounding involvement. You might say: Mono is a community. Java is a corporate process. It's pretty clear which on is going to grab the Linux world.

Enviado por Jod en julio 09, 2004 a las 06:26 PM CEST #

A couple of corrections. I thought I'd said this the first time round, but possibly not. _JCP_ The JCP is not open - you cannot reimplement the Java standard libraries (which is a fairly critical part of making an opensource version) and belong to the JCP. You are FORCED to license the standard libraries in order to belong to the JCP. The gcj hackers want this to change. Also they want to be able to easily license the Test kit, which doesn't allow them to publish results, which is a bit of a pain. _Saving Sun Money_ I don't think we care about saving Sun money, but Sun don't seem to care about some elements of the standard library, which is deeply annoying to people - for example, there are stacks of unfixed bugs in Swing that we keep running into - random ones relating to JSplitPanes not working properly if you initialise them in a certain order. It's very frustrating. However, I do agree with you - I don't see the NEED to opensource Sun's standard libraries. I think that having multiple implementations might well be a good idea, as long as the TCK is freely available. The GCJ people aren't begging Sun to open source their standard library. They want the test suite and that's it (well, I think they'd like Swing to be open-sourced, and I think it would be a good idea, but I think that Sun just won't) May I just point out that Java3D is not open-source at all? The demos are, but the actual implementation is not open-source. Very disingenous. I hated Java at uni, it annoyed me, it felt klunky and strange. At work I've begun to quite enjoy it, thanks to using Eclipse - the first IDE that doesn't make me want to shoot it through the head, hell, the first IDE that really helps me. Eclipse has really pushed Java hard in the open-source world - people see SWT and realise they can build a native app (or not) - suddenly they can ship a .exe that looks and behaves like a Windows/Linux application and it's excited the opensource world. If the classpath hackers continue their work, then we might have native swing too in an executable where we can install it just like any other application. Java does not need to be open-source for this to happen, but the TCK, and fixing the JCP would help

Enviado por Andrew Shuttlewood en julio 10, 2004 a las 03:51 AM CEST #

Hi Antonio, as a kaffe developer, I agree that it would be nice if Sun supported free runtimes in their efforts to be compatible with the specifications. Everyone agrees that WORA is cool & important. But as Andrew points out, noone in the free runtime community wants to get contaminated with SCSLd source code, as that would prevent them from contributing to the free software efforts. So it would be pointless to join a process that requires giving up the right to write free implementations, if you are already writing free implementations. But I assume that JCP is a nice club, if one is a licensor of Sun's implementation. Mind you, the TCK would be nice for free runtime developers to see what one needs to do to improve/reach compatibility with Sun's implementation. Of course, if Sun doesn't want that, that's fine, too. Noone can tell Sun what they should or should not do with their code. If they want WORA to extend to runtimes beyound those based on Sun's code, then liberating the TCK would be quite helpful, though. cheers, dalibor topic, waiting since 2004-04-27 for sun's marketing/licensing to figure out under which terms they want to release the TCK for 1.5. Slow marketers you've got yourself there ;)

Enviado por Dalibor Topic en julio 10, 2004 a las 04:58 PM CEST #

Let me approach this from one (personal) point of view. I'm not saying this is typical of the whole of Java, but it's a case in point... Swing has a bunch of problems. One of them is that its Windows Look & Feel never quite looks and feels exactly like Windows. Admittedly, this is a tough and picky thing to get exactly right. Everyone has a slightly different level of tolerance for the level of inconsistency they'll accept. Typically only pixel geeks will notice the smallest differences. However, enough of these small differences and users will be aware that they're not using a native application, just not sure exactly how they know. For whatever reason, some bugs logged against the Windows look and feel are rejected by Sun [e.g. bug 4818418] even though they are clearly inconsistencies with platform UI. Other bugs get fixed, but the fixes are still not correct (e.g. the menus in the Windows XP look and feel in 5.0, which look a bit more like Windows menus compared to 1.4.2, but are still by no means correct). I'm not sure how many people work on this small part of Swing in Sun, but I do know that a large number of developers in the Java community are perceptive (possibly picky) enough to know that Windows Look & Feel is wrong, and what needs to be done to fix it. The winlaf project (http://winlaf.dev.java.net) brings a bunch of those people together. The trouble is, our fixes will never go back into Swing even if they're right, because Swing is not open source. So why do I mention this? Well... SWT is gaining ground on Swing right now. And a big part of that is because it looks "right". It's harmful to the Java platform as a whole to have two totally different UI APIs and toolkits. Granted, that gives us choice, but it also divides us as a community to a certain extent (c.f. ongoing raging debates in the community about which is best, Swing or SWT).

Enviado por Brian Duff en julio 10, 2004 a las 10:56 PM CEST #

I think you may have missed my point about how being able to test in the wild might have prevented the EJB debacle. The JCP obviously fell down on that one and is only now - we hope - starting to right itself. But anyway, I'm glad you're in favor of opening the specs up more (NDAs and other clauses are a major obstacle). That would basically mean that open-source Java \*will\* happen: given the proper specs, we can make a compatible product.

Enviado por S. Hearn en julio 11, 2004 a las 12:41 AM CEST #

Enviar un comentario:
Los comentarios han sido deshabilitados.
About

swinger

Search

Archives
« julio 2014
lunmarmiéjueviesábdom
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
   
       
Hoy