Moderator & Guest Speaker (SPK): SPK-TomChavez

SPK-TomChavez: Hello, everyone! Since this is a "turn-the-tables" chat where we want to know what you think, I've made this room un-moderated.

uomini: Hi, Tom.

ErikaStuart: Hi Tom!

hobbsa: Hey Tom.

mitchgart: Hi from me too.

SPK-TomChavez: (For those who have never visited before, this means that you can post your responses and comments directly. Let's try to follow some "unofficial" protocol, ending comments with <done> so that others know when they can add a comment. The subject for this week is the "Future of Java." We're already a few weeks into 1997, and if it is anything like 1996 was, it will be a wild ride. So let's start with some of the questions that we posed in the announcement of this forum.

uomini: The JDK seems to be moving along briskly; it's in its 3rd Beta release already. Any idea when FCS will be?

SPK-TomChavez: JDK is nearing the end of its beta cycle, of course, depending on not finding serious bugs. I don't have the JDK product manager here to give me the official date, but it's not months away. We'll have to have the JDK team back! (And the Java Server team, and the ....) My turn for a question. What would you like to see happen on the Java front this year? (One at a time...)

uomini: As for future releases of Java, I'd love to see an HTMLCanvas class, which would let developers roll-their-own browsers. Think of the possibilities...

jimheintz: Anyone there?

hobbsa: When (if ever) will Java support MDI applications?

spietrowicz: I'd like to see the Java Media classes start to appear ASAP. I have people working on things that could use the JM classes now, but since they're not out yet we have to "roll our own."

mitchgart: Boy, I'm sure finding a lot of bugs in JDK 1.1, new ones that weren't in 1.0.2. I've been submitting lots of reports. What I'd like to see is a really good job at fixing the existing bugs in AWT and other API packages, before adding zillions of new features. I'd like to see better support by Sun for existing features.

uomini: Amen!

jriedese: I like the beginnings of Java Spaces. I'd like to see that pushed and start seeing research akin to David Gelernter's "Mirror Worlds" start taking place.

SPK-TomChavez: (I haven't heard of Mirror Worlds but maybe the Java Spaces team has... I've seen lots of requests for all of the above. Bugs should be addressed before JDK 1.1 goes final.

hobbsa: How about shared variables that are used throughout a group of instantiated objects of similar type.

SPK-TomChavez: User interfaces is a good question. Now that Java is coming out in so many places, the idea of the native look and feel comes into question. What about devices that have no native look and feel? (No need to answer that one now!) Media classes (spec, implementation, ...) is a popular question. I'm asking the Product Manager to keep us up-to-date on all of the activities and work. You've asked some specific feature questions, such as the html canvas. Others? AWT? Security? Networking? Commerce? What should we do for JDK 1.2, 2.0, ...?

hobbsa: When will Java offer Toolbar Buttons or something similar

spietrowicz: Isn't the HTML canvas question being addressed by HotJava Views?

jriedese: Taking ideas from EULisp and the Newton OS, what about defining Java in a layered approach so that it can be easily embedded in many different devices (I can envision many devices that have no GUI, for example). Newton OS seems quite embeddable, and the EULisp standardization was
designed to be very layered. My concern for embedding java in different devices is obviously a problem of potential size - but with Java on a chip - perhaps this isn't a problem.

uomini: I'd like to see Sun firm up the business foundation. By that I mean cutting deals with major HW/SW providers, the goal being to entrench Java as much as possible.<done>

SPK-TomChavez: (Interesting -- a business request... I hadn't gotten to that yet! Let's hold on those for a minute...) What about 3D? Or prioritizing the media classes, speech, ....

mitchgart: Tom said "user interfaces is a good question." Well, what about fixing bugs in TextArea and FileDialog and other basic AWT things? How can one write a decent user interface if the basics like these things are broken? Seems to me like Sun is a little too focused on the long term glamorous big-buzzword research-y stuff and isn't willing to work hard on fixing the basics. 3D, speech, great, but first fix the bugs in the basic classes.

hobbsa: I agree. Make Java a viable business language and then work on the fancy stuff.

jriedese: 3D is important. I mean as soon as we start implementing some of the cyberpunk literature and have real 3D devices to engage with ;-). (It could happen :-|) I agree with getting the foundation working *well* though (first).

uomini: I haven't encountered many bugs in Java, but if it's as broken as you say, I second the motion.

SPK-TomChavez: (I hope the bugs you're finding in TextArea and FileDialog are being addressed in 1.1. Are they in the b3?)

spietrowicz: My priorities for the other classes? Java Media, 3D API, JM Audio, the Speech API, JM Video. Another thing I'd love to see is the ability to save image files in their native formats.

SPK-TomChavez: Even I agree about fixing basic area problems. Stable foundation and all...

hobbsa: B3 finally fixed a problem I was having with MenuShortcut, so Sun is working on problems, but I feel they could work harder on some of the basics instead of the glitz.

mitchgart: The bugs I'm talking about are in B2. I'm just downloading B3 now in a separate window, will try it soon.

SPK-TomChavez: The good news is that we have multiple teams working on issues at the same time. Some fix bugs while others work on new features. So we can do both.

Keep the bug reports coming in. They go into our bug database and that gets them resolved. If they're not in the database, they don't exist. So you have to report them! And thanks for those of you who have and do.

spietrowicz: Someone asked privately what I meant by "native formats"... I meant GIF, JPEG and PNG.

mitchgart: I keep sending bug reports but I don't see them in the online lists Sun publishes.

SPK-TomChavez: Thanks for the clarification. I'll pass that on. I'm not sure if they list EACH and EVERY bug that is fixed or just the major ones.

jriedese: Could be a business comment...I'm concerned that Sun is spreading themselves too thin by doing way too much and not focusing well. At the same time I understand the need to get Java entranched - I mean we all want Java to be there for the long run. Is it possible to throw more resources at it?

SPK-TomChavez: Sun is a BIG company and IS throwing lots of resources at it. JavaSoft is bursting at the seams, and we're just one of Sun's divisions working on Java technologies. And we'll keep adding more. So the short answer is yes.

mitchgart: "The future of Java" is our subject. I think maybe Sun is taking the attitude that what Sun develops is equal to "the future of Java." If it's developed by Sun, it counts, otherwise not. I think that's wrong. I think if you provide the basics well, third parties can create a lot of the glitzy things. "Yes!" to the comment about focus that jriedese just made.

SPK-TomChavez: Let's move forward a few years to the year 2000. Comments on this can be about the Java business or the Java technologies. I'd like to see Java as a multi-billion dollar dollar industry.

uomini: Yeah, I'd like to see Java make it BIG in consumer devices. If it succeeds, then it will become mainstream.

hobbsa: I would like to see Java developers at about 200$/hour :)

SPK-TomChavez: In the year 2000, I would think $300-400 an hour would be reasonable!

thevoid: When will Sun give Java to a standards group (into the public domain)?

uomini: Might be minimum wage, for all we know ;-)

SPK-TomChavez: Sun has hosted standards meetings already. I've read that in the press! I don't know the details.

thevoid: Sun is still keeping the standard inside.

SPK-TomChavez: Java in consumer devices is an interesting thread. What are your thoughts of which devices? VCR? Telephone? Cell phone? Calculator?

mitchgart: Year 2000: Java and Windows are neck-to-neck in a race for mindshare
as the main development platform. The Java platform is the language
+ AWT + Beans + a few other packages. There is a big thriving third
party software market. There is a standards group and a validation test
suite for each major piece of Java software. Sun software has to pass
the test suite along with third party software. :-)

spietrowicz: Tom... PDAs first. Hook them up with a wireless network, and you'd have some interesting applications.

uomini: One idea I suggested to Scott a few weeks ago is for Sun to get into the ISP business, for the purpose of providing Java-based hand-held devices an Internet connection. This could be very lucrative, and is necessary that there be world-wide access providers in order for Internet-based PDAs to succeed.

SPK-TomChavez: Interesting. Out here we have wireless 28.8k access already.

jriedese: I believe that the FCC has just opened up some new 100 MHz wide bands in the GHz area. These are unlicensed and will allow 1.544 Mbit transfer over a few miles line-of-sight. This should allow networking to really take off. I'd like to see Java take off with that.

SPK-TomChavez: Good suggestions for the JavaOS licensees.

thevoid: The AWT is a disaster for PDAs, how about a decent UI system that can scale devices?

SPK-TomChavez: Something that would work for LCD displays, ...right?

thevoid: The UI needs to go from 1-bit devices (flashing LED) all the way up to multimedia--I want to check my e-mail, the bit is useful, then a one liner, etc..., all from the same application running on different devices.

SPK-TomChavez: With multiple input device support from push button to handwriting recognition!

jriedese: ... and body language recognition ;-)

spietrowicz: I've seen wireless ethernet around here through a PCMCIA card. A PDA with
one of those would be nice to have for meetings. When people move out
of the office, they could switch to the more costly wireless networks. The
nice thing with the local wireless network is that after the initial cash layout
the costs aren't per packet or time based.

SPK-TomChavez: :-) What about the business? We covered consulting fees... Do you all want to make money, and if so, how?

mitchgart: In the discussion forum applet there's the "compose multiline message"
button, which sends a bunch of lines together.

uomini: I don't want to make money: I already have enough :-)

SPK-TomChavez: Ok, no business insights yet...

uomini: Hold it...

SPK-TomChavez: Online sales of software is an idea that comes up again and again.

uomini: I see many avenues to make big bucks via Java, other than consulting...

thevoid: How about leasing software by the second? (IBM was successful in the past--only yearly).

jriedese: This internet/web revolution has created an entirely new industry.
Demand, and therefore money, isn't going to be a problem.
But what's nice is that this new industry is only now being defined.
Meaning that other companies (M$) aren't in this driver seat. I think the opportunities are only as large as the imagination (and the cyperpunk novels ;-)).

uomini: For example, Java/Internet classes for businessmen, SW toolboxes with simple yet compelling applets, etc.

SPK-TomChavez: Leasing by the second. Wow, that'll get me in and out quickly! Custom software packages?

thevoid: So there can be an advantage to a slow interpreter.

uomini: Not custom, just a small set of very useful things for businesses, e.g., animation applets, live forms, etc.

mitchgart: I think in the future we'll see people writing programs in Java and
shipping two versions: a portable bytecode version and a Windows
executable.

thevoid: Hopefully they will only have to "ship" the windows version The Java can dynamically update!

SPK-TomChavez: Why two versions?

uomini: I think we'll see native code binaries for apps, and not just Windows exes.

thevoid: Currently a single version does not look sexy enough on Windows (a popular OS). And to keep up with MS, it needs to be friendly with the other OS apps.

uomini: I'm looking for speed, and right now that means either executing under JavaOS or compiling to native code.

mitchgart: The portable version will run everywhere, the Windows .exe version
will run faster on Windows than the bytecodes.

SPK-TomChavez: What about JITs that save the native code for close to native speed? Almost as fast as native compile?

uomini: Sure. When will the Sun JIT be out? The first Beta was very buggy.

hobbsa: Why doesn't Sun introduce JIT for most of the major platforms?

thevoid: I think you are overestimating the need for speed, big picture ease-of-use is critical.

uomini: The Symantec and MS JITs are too slow.

SPK-TomChavez: JavaSoft creates the reference implementation, the licensees create the ports and improve the performance.

uomini: Oh, yeah? Just try using a Java-based newsreader and displaying 22,000 groups...

mitchgart: Interpreted bytecodes are 10 or more times slower than native programs.
Using JIT, the ratio is only about 2-1. That's a big improvement but still two times slower.

thevoid: Besides, much of the speed pain is the AWT out of sync with the OS. [22,000...] Okay, but the lack of confidence in the job ever finishing is much of the problem.

uomini: I disagree. The JITs I've used are maybe twice as fast as the interpreter...

uomini: In my case, the AWT is not the only part of the problem.

thevoid: Explain?

jriedese: I have to say that I find it very interesting that one of the big arguments against LISP has traditionally been speed (interpreter). It's interesting how that isn't levied against Java in nearly the same way (or nearly as much, overall). However, LISP has been dynamically compiling methods to native code for years. Seems like it's just a matter of time to get the JIT's to work well.

SPK-TomChavez: I agree.

thevoid: Of course, that's not the problem.

SPK-TomChavez: In addition to speed, the platform needs more platform features.

thevoid: What do you mean?

SPK-TomChavez: I've seen requests for more filling out of the feature sets, so that a Java application is not limited by being 100% Java vs. native. (Looking at the clock.. we're running out of time. I did want to get to our last two questions)

thevoid: Cheers.

SPK-TomChavez: What would you do if you were Sun's Java product manager, and what would you do if you were the Worldwide Java Czar, and could do anything that you wanted with the technology?

uomini: I'd promote my idea of getting Sun into the ISP business for PDAs.

jriedese: (I think there's a bug in the multiline composition - every other one
of my messages make it, the others don't.)

SPK-TomChavez: Okay, I'll start. As Worldwide Czar, I would make Java world-ready with each release. Not later, the same time.

mitchgart: What would I do ...Like I said before, fix the bugs first. I mean, it's great if three different companies make three different speech recognition packages and the market chooses the best one. But it doesn't work for three different companies to make their own AWT.TextArea. Sun/Javasoft have to
do a good job on getting the basics right, otherwise there's no foundation.

spietrowicz: I'd push PDA vendors to get Java into their products so developers could start working on it.

SPK-TomChavez: And release the classes for PDA development, too.

uomini: Yes!

spietrowicz: yup.

uomini: If some world-wide ISP were to team up with PDA manufacturers, that would be an incentive for Java-based PDAs.

jriedese: Instead of simply going with the flow, especially with respect to GUI, define totally new concepts and build a much better idea for GUIs (they've kind of not progressed well, in my opinion).

spietrowicz: Not necessarily... I think that if the PDA stuff were put in, the ISPs would follow in due course.

SPK-TomChavez: It's noon here in California and that means time to close another chat.

uomini: Yeah, but the consumers won't buy the PDAs if they have to deal with the ISPs themselves. The key here is bundling worldwide Internet access with the PDAs.

mitchgart: Bye.

SPK-TomChavez: Thanks everyone for coming. The log of this chat will be sent around JavaSoft as input.

spietrowicz: Thanks Tom.

jriedese: Thanks!

uomini: Bye!

SPK-TomChavez: Bye, all.