Sun Stands Firm

Q & A With Sun at Today's Press Teleconference

PALO ALTO, Calif -- September 22, 1997 -- Following today's press teleconference, Sun took questions from the industry press. The questions and answers are transcribed here.

Sun Microsystems Speakers:

Jim Mitchell, VP of Technology and Architecture
Alan Baratz, President, JavaSoft Business Unit
George Paolini, Director of Corporate Marketing, JavaSoft Business Unit
Lisa Poulson, Sr. Manager of Public Relations, JavaSoft Business Unit

Operator: Ladies and gentlemen, we will now begin the question and answer session. Mr. Patience, Computer Wire, go ahead with your question.

Q: Jim, what do you mean when you say you will submit the spec for the Java platform? What elements do you consider comply to Java's platform?

Mitchell: Yes, Nick, that's pretty easy. They are the Java Virtual Machine specification, the Java Language specification and the specification for the API for the class libraries that are in the core platform. Right now those correspond to four books in the Addison Wesley series. The APIs are in volume I and volume II. That's what it is.

Q: Right. And that will be what comprises the specification for Java?

Mitchell: That's right. It's what's called a multi-section specification in the standards parlance, but it will have chapters that correspond to those three major areas.

Operator: Torston Busse, from IDG News Service.

Q: Good morning. I wasn't really entirely clear how the submissions you're making by the end of today differ from your first submission. Can you really expand a little more on how you have altered your submission compared to the one that was voted on in July?

Mitchell: First of all, we're not doing another submission. The process with ISO is a two-step process. You go out to get comments and an initial indication of a vote, then you respond to those comments. We're responding to those comments. We're not starting over again in any sense of doing a resubmission. Secondly, in those three major areas, the one about scope was simply a question that they wanted to make it clear and we've clarified that. It was what we always intended to do. On maintenance, this has taken, I think on both sides, understanding of what the issues around maintenance are and finding a middle road that both meets JTC1's written policies and procedures, but allows us to keep involving the Java industry in the maintenance and evolution of Java. In lots of conversations with JTC1, as well as our developers and licensees, we found what we think is a good middle road where both groups will be satisfied.

Lastly, on the intellectual property rights, that also was a clarification with one exception. That is, there was one comment that asked that we actually give our brand away. We obviously cannot do that, and we're not doing that. But we have made it clear that the Java name can be associated with the standard, and our normal guidelines for using the Java name apply to anybody who does an implementation from those specifications. If I went outside and created a clean room implementation of those specifications, I might call it something like Cappuccino. I could say this is Cappuccino, an implementation of the ISO specification for the Java platform. That's the sort of thing that is allowed by this.

Q: Well, maybe I wasn't clear with my question. On the second part, the maintenance element, when we talked previously you said you made some adjustments to it and you ran it by ISO members and they seemed pretty happy with your response. Maybe you can explain exactly what this middle ground is that you found?

Mitchell: The middle ground is that there will be a working group within ISO. That's our proposal. That's what they would want to set up. To hold the specification, we'll of course be involved in that working group and they will go through ISO's normal processes for doing maintenance. At the same time, we will still go to our developers and listen to everyone who says I found an ambiguity or a bug in one of the specifications that need to be cleared up. Just really involving both bodies.

Operator: David Carr, Web Week.

Q: This also, I guess, concerns the working group and the distinction between that and your own process. What I'm hearing is that Sun continues with the process that's been followed to date. It comes up with a recommendation for what changes should be made, but that doesn't become official until it's ratified by this working group.

Poulson: I think part of the reason this is becoming a big issue is because people are thinking maintenance means a lot more than it does. Maintenance just means fixing bugs and fixing ambiguities. It does not mean adding new things to the Java platform at all. So it is simply the specifications you've already submitted, we found a bug or we found some warning that needs to be changed. So, it's housekeeping.

Q: All right. When you go from Java 1.2 to 1.3 or whatever it winds up being, how does that happen within the standards process?

Poulson: It has nothing to do with maintenance at all. Jim, can you explain that briefly?

Mitchell: Lisa is exactly right. Maintenance is fixing these bugs and everything ,and there is a process in JTC1 to go through that and put even those things up for voting and comment. But if Java changes in some substantial way what we would expect to do is a new submission of the Java platform specs. Now, if the language in the VM, for example, didn't change then no one would be fussing about those but they would certainly want to look at a new API and be able to comment on it in its entirety.

Poulson: So changes to the platform are new submissions. They do not go through the maintenance process.

Baratz: If I can just take some of the hypothetical--

Poulson: We've got time schedules from 9:30 to 11 for one on ones and we'd be happy to talk to and I'll have somebody call you and we'll spend some time with you on the phone.

Q: Thank you.

Operator: Mark Schlak from Byte Magazine.

Q: As I understand, Jim, this process that you're laying out, if six months or a year from now, you and IBM or you and Netscape have a substantial disagreement about how to add something to the Java platform or how to develop something within the Java platform, in fact, you have not established an open process to deal with that. You would essentially listen to their concerns and make your decision as such.

Poulson: I think we're having a really huge communication breakdown here. The process for how the Java platform evolves has not changed at all and is exactly the same as it has been for the last two years, and it involves many companies beyond IBM and Netscape and Jim defined the process 10 minutes ago on the phone--Jim, could you repeat...?

Mitchell: First of all, we don't listen to just one company.

Q: Any one of your partners. I didn't mean to pick on IBM or Netscape.

Mitchell: It can't be just one. And that includes us. The primary value proposition of Java is Write Once, Run Anywhere. When the entire computer industry shifts it in order to keep write one run anywhere we have to get consensus around changes. That's why we run as open a process as we do that involves all of our partners, all of the people on the Web who care about Java--both developers and end users. We can no more veto or do something with just one company--if we do that we have killed that right to run anywhere, so our process is open because it has to be. That's in enlightened self interest. It's not altruism. It's not something we do just for show. It's something we do because it's part of making Java successful. We will continue to do that. It involves licensees and as I said the entire Java community. They make real contributions that make fundamental architectural changes to things like JavaBeans, for example.

Q: I have one follow up question. What kind of revenues does Sun realize from licensing Java?

Baratz: Sun generally does not break down its revenue by product or by operating company.

Poulson: We're not allowed to answer that question.

Operator: Michael Hiltzik, Los Angeles Times.

Q: Thanks, Alan and Jim. One of the elements of Microsoft's campaign as it were against Java is a fairly comprehensive attack on the quality of the language itself. And on its potential and it's robustness as a possible substitute platform. Recently I heard a Microsoft executive say that the Write Once, Run Anywhere paradigm is a fraud--I think those were his words--because 10% to 20% of an application has to be rewritten before it can be ported over to another platform. Can you comment on these challenges to Java's basic potential, in terms of its scalability, its robustness, and so on?

Baratz: Sure. First of all, I think the key comment in what you just said was Microsoft's attack on Java. Let's just understand where we are. We're talking about a monopoly, Microsoft, that is trying to preserve their monopoly and will say nothing, will take no action other than that which preserves their monopoly. So, anything Microsoft says, anything Microsoft does with respect to Java needs to be understood in that context, a monopoly that is trying to protect its monopolistic practices. That having been said, what we need to do is break down some of the concerns that have been raised by this monopoly around the Java platform. The one they've repeated most often lately is write one, run anywhere is not in their view as we say it is. The truth of the matter is that we are now seeing the truth beginning to come out as we are beginning to see articles written on Java applications that have been certified as 100% Pure Java and the extent to which they actually do run anywhere. So, for example, I would point you to a September 15 article in InfoWorld where they were the first to actually test a set of applications that had been certified as 100% Pure Java and guess what they learned? They learned that in fact those applications did run and run well on all the platforms that they tested it on.

Now, it's true that if you write an application using the Java language but the underlying OS specific calls are embedded, either on purpose or because you're unaware that you're using the underlying system call, then those applications will not run anywhere. They will be tied to the underlying operating system platform. The problem that we've had up until now is that the testing that's been done on the applications has been done on application for which the developers have included system specific calls either knowingly or unknowingly. And as a result the applications are locked to a particular platform. The reason why we launched the 100% Pure Java certification and branding program several months ago was to help developers understand how to write applications that don't include those proprietary platform lockins. And to provide them an independent testing and certification entity to validate, to verify for them that the application has been written as 100% Pure Java and does not include any proprietary system login. The tests that you want to do are on the applications that have been written as 100% Pure Java. That's where you'll know whether or not the Java platform is truly Write Once, Run Anywhere or not. The fact that as the tests are being done now on the 100% Pure Java aps we're finding that the value proposition of Write Once, Run Anywhere is in fact intact.

Baratz: The other thing I'd like to add is that the primary reason the organization of the Java Lobby was created was because those developers believe in Write Once, Run Anywhere--you can read it in the editorial on their website at In less than two weeks they've already accumulated 1,200 developers as members of the Java Lobby who agree with this, so there is also belief in the developer community that this is very important.

Operator: Joe Barr, DweebSpeak.

Q: In regard to the war with Microsoft, do you expect Microsoft to remove Java support from their browser?

Baratz: I don't know how to speculate on what Microsoft might or might not do. At this point in time Microsoft is a licensee. Microsoft is delivering Java within their products. The current Java in these FCS products is compliant with our reference implementation, passes our tests. I can't really speculate on what might happen in the future.

Paolini: I want to emphasize one point you just made. This really is not a war with Microsoft from Sun's perspective, but it is Microsoft's reaction to where the industry is moving. I think that's a good insight that you just had.

Operator: Maureen O'Gara from Client Service News.

Q: Have you guys done any soundings amongst the national bodies of JTC1 that would suggest that the replies you are going to post today would be sufficient to carry the day? And if it's not--one of the objections a lot of those guys made was to the notion of Sun or any commercial entity being a PAS applicant in and of itself. What are you going to do about it?

Mitchell: Two things. One is yes at the JTC1 plenary held the week before last in Ottawa we held an unofficial meeting outside that plenary and invited members of the national bodies to come. It ended up being a standing room only audience. We explained all of these same responses to people there. I thought that actually went very well. There was a lot of head nodding in the room. I of course talked to as many of the delegates as I could one-on-one also. I would say the only people really strongly not convinced were Microsoft and Intel. So we have done that.

The issue about a single company being a PAS submitter is such a non-issue. The PAS process was designed to allow even single companies that JTC1 has actually issued a letter saying we don't even need to respond to those few comments. And there were only a few of them that said that. In our response we don't actually deal with that issue, we just include the letter from JTC1 that says that's an issue for JTC1, not about our application.

Q: Your expectation is that this will carry.

Mitchell: My expectation is that this will carry.

Operator: Hiawatha Bray, Boston Globe.

Q: I have a simple question. I think it cuts to the heart of the matter. What if you lose? What if JTC1 goes thumbs down? What do you do to drive the Java platform forward, and can you do it regardless of how that comes out?

Baratz: The answer is we're going to continue doing exactly what we've been doing up to this point in time. We have a well defined, open industry participative process for evolving the platform. We have a strong branding program around the Java brand for our implementations of Java. That brand is targeted to preserving Write Once, Run Anywhere; Safe Network Delivery, Supercomputer to SmartCard Scalability. None of that will change.

Q: There have been a few criticisms leveled at Sun for much the same thing that Microsoft is being criticized for. People are saying you guys are trying to hand onto Java precisely so that you can become a dominant operating system the same way they did with Windows. What do you say to that?

Baratz: I think there is a huge difference between the way Sun operates and the way Microsoft operates. Sun operates through an open industry participative process that yields open specifications with free access to the intellectual property needed to implement against those specifications and then separate from that, Sun delivers its own products written to those specifications into the marketplace. Those products need space in which we derive revenue value from the technologies. Microsoft, on the other hand, pursues a very closed proprietary process with respect to both the specification of its technologies as well as the implementation of its technologies. The process that Sun has been running, is running, and will continue to keep running, is that open industry participative process. It will continue to yield open specifications for the APIs, it will continue to be delivered with free rights to the intellectual property needed to implement against those specifications. And then Sun will continue to deliver its own branded product against those specifications.

I would urge you to go to and see if you can find the API specifications that make up Windows. See if you can find the intellectual property agreement that would allow you to implement to those specifications. You won't find it. You will find it at Sun's website. You will find it in many, many publications. Back to your original question, the answer is we very much would like for JTC1 to approve our PAS submitted application. We are hopeful. We are confident that they will. If they don't, we will just continue doing business as we have been to date.

Operator: David Folger, Meta Group.

Q: Thank you. We've heard a lot about industry participation in the process of defining the specifications for Java, but one of the other questions is to what extent are you getting end user input, particularly from the Global 2000 IT organizations? Many of them are thinking of planning to using Java in a key way within their organizations. And, by the way, Microsoft does try to get input from them even though they don't perhaps get as much as you guys would like from the other vendors in the computer industry.

Mitchell: In the last stage of our process, we opened up specifications to the entire Web. We don't just say, help us fix typos. We interact on that as major issues are found to revise the specification thing and give anyone on the web another chance. That includes not just developers producing applications for sale but also developers inside of enterprises, and it includes anyone who wants to. We don't require that they be a developer; we get comments from users, people who have interests from university and research perspectives. We get them from everybody. An example is when we first put out the JavaBean specification. In very short order we had 40,000 downloads of that specification. We iterated it three times and two fundamental architectural changes were made by people out on the web, telling us things they thought should be changed. Many of them were developers, of course, but some of them were in fact end users and university researchers.

Q: Do you have any plans to involve committees of senior IT professionals from end user companies in the specification process, perhaps on a little bit more formal basis?

Mitchell: I don't think we need to do that. The computer industry is pretty much our licensees, so they already are doing that. And for the people who are not already licensees, there is the web-wide participation. I know we get input from folks inside the enterprise, even from CIOs, and so I think the process already includes them and keeps a lot of speed.

Baratz: I'd just like to add a comment. More recently we've launched an online service called the Java Developer Connection. That service has been up and running for a couple of months now and already we have over a 100,000 developers registered on that service. If you go to that website you get all kinds of information. You participate in chat sessions with the people who have developed and are developing the Java specifications and the Java implementations. You get early access, and it's a vehicle through which you can provide feedback to us on the specifications, on the implementation in addition to participating and identifying bugs in the implementations and helping to fix those bugs. So here is a community 100,000 strong of developers that span from the small startup companies to the large third-party ISV, to the corporate IS organizations. We have developers from all of those organizations that are involved.

I should also point out that as we are now moving to deliver more focused enterprised products. For example, we just launched the Java Web Server, which is an HTTP server written entirely in Java for the enterprise environment. Or the Java Webtop, which is a new Java desktop environment written entirely in Java--e-mail, calendaring, directory services. We call it HotJava Views, targeted to the enterprise. As we are moving more aggressively ourselves for products into the Enterprise, we are engaged more and more heavily with the CIOs and with the business managers within the IS organization and we are now gathering more and more feedback from that group of individuals as well.

Operator: Cheryl Fenelee, Miller Freeman.

Q: I'm wondering, with all these avenues for feedback, through the web, through all the different companies you work with, how do you choose which suggestions to implement first?

Mitchell: That's pretty simple. When we start seeing a lot of people asking for the same thing, then we better pay attention to it. With that many developers, with this many licensees, when it isn't one single company that comes up and says we think you should do some new thing "x"--it's almost always all of the companies in a particular industry segment who think we should do something. You don't have to be smart to figure this out once they all start saying you should do something in the area of 2D graphics. We get it, and then we start moving on it and involve the experts out of some of those same companies who have been bringing it to our attention.

Operator: Alex Wolfe, EE Times.

Q: Given that the platform independence or Write Once, Run Anywhere is so important to the success of Java, how do you keep Java from having the Unix problem? When people start adding their own extensions byte codes? Do you foresee having a Unix-type problem which breaks the platform independence of the language?

Mitchell: You've got it right. That's it exactly why we run the process we do, so that we can have it be open--which Unix was--but avoid the Unix fragmentation and keep it unified, so Write Once, Run Anywhere keeps working. Everything we do derives from Write Once, Run Anywhere and so we have to keep it unified. That's why we build consensus among our licensees, which is basically the computer industry, as well as all the developers out there on the Web. To prevent that kind of fragmentation.

Q: What happens if people do clean room--like you said before they call it Cappuccino, but it's non-standard and it has extensions which aren't supported?

Paolini: One of the points that Alan made earlier on was that we have invested heavily in the Java brand. A major distinction between Unix and Java, technology aside, is the power of the Java brand. We have been very effective in using the brand to essentially keep the industry on one platform, one Java platform. We have done this by using the brand to certify and allow the use of the brand in various substantiations for those who comply. If you don't comply, you will lose the right to use the brand. So far that has worked extremely well. I think that's a major distinction in the marketing side of where Java is versus where Unix is and was.

Operator: Debbie Gage, Computer Reseller News.

Q: If this does not go through, what impact might that have on the power of the Java brand? I'm wondering if you could answer that and spell out more why Microsoft and Intel are not convinced by these responses?

Baratz: First of all we think it will have no impact on the power of the Java brand. The Java brand is what it is because of all of the work that Sun and its partners have put into the technology and into the promoting of the brand and into insuring that the brand continues to stand for Write Once, Run Anywhere, network safety and scalability. None of that will change. And all of that was done totally independent of ISO or the ISO PAS submission or any other formal standard bodies activities. While we're hopeful and confident that the PAS submission will be accepted, if it is not none of that will change. We will continue moving forward exactly as we have been and be believe that the value in the technology and in the brand will continue to grow as it has to this point in time.

I don't recall your second question?

Q: Why is is that Microsoft and Intel won't accept these responses? Jim said they were the only ones?

Baratz: It goes back to remembering that Microsoft is a monopoly and Microsoft will not do anything that jeopardizes their monopoly. Microsoft is interested in one thing and one thing only, and that is fragmenting Java and destroying the value in the brand. Microsoft understands that when the brand is managed by a business, the value proposition is kept preserved and when the brand is turned over to a standards organization, they don't have the wherewithal to manage it to the value proposition that's required. So all Microsoft cares about in the context of this submission is destroying the unification and destroying the value proposition in the brand.

Paolini: I'd like to take a little bit different cut at it which is that what Microsoft would really love to see is its ability to take the brand and be able to use the brand to mean other things. Again, Sun has been very effective in cooperation with the majority--99% of the computer industry--in ensuring that the brand means the three things that Alan said. Write Once, Run Anywhere, Safe Network Delivery and Scalability. Microsoft would love to be able to take that apart and especially they would love to annihilate the abilities, the portability aspect of Java. The only way they can do that is by pulling the brand apart and we will never let that happen.

Operator: Carl Braunstein from the Robert Francis Group.

Q: At this point in time Sun is applying as a publicly available submitter. Most of those are standards bodies. Do you see that having any impact, does that give you the ability to do all of the updates without any control by the ISO organization and do you see that as a problem?

Mitchell: Most of the other PAS submitters are a mixture of standard developing organizations like ECMA and Consortia like OMG and so on. So they're not all standards developing organizations at all. Because of the process we run and the openness of that process and how it involves basically the entire industry and all the developers and users around the world, in terms of evolving the platform, in terms of getting input on bugs and ambiguities and so on I think there is absolutely no problem with making this work. Also bugs and ambiguities and so on can be entered through any process the JTC1 wants to set up so that they'll get considered as well. I don't see any problem with this. We run an open process that is more open than any standards developing organization or consortium in the world, at the moment. In fact some of them have been picking up our White Paper and considering using some of these same mechanisms and using the Web to become more open.

Operator: Busse, from IDG News Services.

Q: Maybe this is obvious, but could you, after you reiterated several times today how open the process is and how well it works, why are you doing this? What's the advantage of going through ISO?

Poulson: Let us take this up with you off line because Jim would be happy to go over it with you again.

Paolini: Let me give a quick response to that. What seems counterintuitive to Microsoft is that the open process makes good business sense. We grow the entire industry, not just one company and by growing the industry we grow the company as well. That's Sun's perspective. That's what Sun has done for 15 years, starting with NFS back in 1982, we had developed open technologies that we have put out into the marketplace and we have benefited from those technologies as well. That's the real answer. It just makes good business sense.

Operator: Nick Patience, Computer Wire.

Q: I understand the difference between maintenance and pushing the spec forward. What I don't understand, when you make the next spec on 1.1.4 or whatever and you say this is the ISO specification for Java, does that actually go through any other processes?

Mitchell: The one advantage to that is any submission we do goes through voting and commenting by all the national bodies. There is a real checks and balances. There is no rubber stamping going on here with the process. That's the normal PAS process.

Q: That's all the committees, actual PAS applications gone through?

Mitchell: All the same national bodies get to comment and vote.

Q: Don't you move this spec on about every three months? Is that feasible?

Mitchell: No, when you say something like 1.1.4, that doesn't have new APIs in it, that just has improvements in performance and bug fixes and so on. When you see something 1.2 versus 1.1 that may involve some additional APIs. That doesn't move as fast as the other things, and of course, in the early days of Java it evolved very, very quickly. That's beginning to slow down. That's another reason why now isn't an appropriate time to begin standardizing it.

Q: Just when you change the APIs to JVM--

Poulson: We'll take with you off line. We only have about 30 seconds left. Alan?

Baratz: All I'd like to say is that we believe that there is value in becoming a PAS submitter. We believe there is value in becoming a PAS submitter because as George says, it grows the industry for all involved and provides another level of review and in some cases there are governmental bodies that would prefer to buy de jure standards so there is some value. However, what we want to be very clear about is the fact that while we're hopeful and confident that the PAS application will be accepted, if it is not we will not change the way we operate. We will continue to drive the same open industry participative process that we have up until and we will continue to drive Java to be the de facto standard. We will just simply not take the next step to turn it into a de jure standard. With that, we'd like to thank you all for having joined us today.

Poulson: We will have the ISO comments up on the web site by the close of business today. We will be posting either a summary of this conference call or a transcript and we will have this telephone call available for replay starting today at 10:30 until the 24th. The replay number is 1- 800-633-8284. The reservation number for the replay is 3189965 and we will put this on the web site. Thank you very much everyone for taking the hour to talk to us and we'll talk to you again soon.