Calvin Austin's Blog

JavaOne - Hits and misses

I really enjoy JavaOne, I've been to pretty much every show since the first JavaOne in 1996. This year I think JavaOne was an eye-opener on many levels, the Java ecosystem is very much alive. The pavilion floor was full of vendors, there were lines for the sessions and even back to lines to the restrooms.

I met some folks who hadn't been to JavaOne since Java 5 came out so it was great to see renewed interest in the platform. Its almost like an industry secret that Java is still such a popular platform.

My hits for the conference were as follows:

High attendance by the Java developer community, many sessions were full
Organization was very professional, not that I enjoyed being scanned everywhere but it worked
Meeting with fellow Java champions and developers

My misses:

Lack of visibility about Java 7, lack of communication about JavaFX. Java developers are patient but don't push them too far
Some sessions needed to be in larger rooms, or simulcast. I understand the cost issue though
wireless network was strained, both for the keynotes demoes unfortunately and around the moscone. The alumni network access was fantastic though!

Applets re-birth, what happened?

A Potted History of Applets

Dion was privy to some news about the latest news on Java Applets. But why a re-birth of applets and how did they get from front page news to an industry footnote?

My first experience of applets was from WebRunner, the Java based browser which spread through Sun like wildfire. I was working for Sun in Europe in those days and it was such a step up in ease of use for anyone who had to mess with motif or X before. In addition to webrunner, was the standalone appletviewer tool which allowed you to keep up with the ever changing api.

However applets were dependent on Motif for Unix, something that didn't change until JDK 5 and Sun didn't have a lot of development history with Windows either. The apis was functional but new, there were many early issues with modal dialogs and later motif related drag and drop issues in 1.2

The big push was for Java 1.0 and of course the adoption by netscape, however deployment of 1.1 applets was immediately affected by the pace of netscape development and the infamous reverse dns lookup feature. For applet developers it meant that many complex early applets would 'break' because they were too new or fell foul of corporate firewalls.

At the same time there was a push to JavaBeans and the getter/setter pattern was applied to awt in a frenzied rush. This resulted in the large batch of deprecated methods event from 1.0 that many Java developers still lament about today.

However momentum was building, Microsoft was a licensee by this stage and Project Swing was the next big thing. Anyone who worked on awt was immediately moved to Project Swing. Swing was very comprehensive, but obviously larger. It introduced the new event model and brought many other library features. It also required JDK 1.1 and didn't become part of the platform until J2SE 1.2.

By now the hotjava browser was losing adoption and Java was totally dependent on the browser plugin apis. Displaying the output of a separate process into a browser as a plugin does was still an impressive technical feat but consumers are more interested in it just working. What was really needed were the plugin and browser teams to work as one but due to many very public reasons that never happened.

So could applets rise again, of course. It requires the browser teams to be onboard and a significant investment in engineering and testing

As with any re-collection I've tried to summarize many events to the best of my ability so any errors are not intentional and corrections are welcome

javascript remote diff/merge tool

There are certain computer tasks that I sometimes wonder if there is a better way of doing them. I mean computers are there to save time, not generate work. One such task is remote file diffs, especially on our webfarm. When deploying updates to our developer site there are often files flagged that were not supposed to have changed.

My colleague Nimish has created a tool that can do a remote diff and merge and display the results in your local file browser. Apart from some python server hooks the rest is written in javascript. What was always a clumsy and error prone task I can now do locally as follows

jsblend.py scp://user@server:/opt/lib/file.pm /opt/lib/file.pm

and see the comparison in a web page.

For more information on JSBlend project and if you want to join check out his blog entry

Pavarotti and I - a true hero

I was very fortunate and lucky to meet Pavarotti in 1982. Its an easy date to remember as it was at a recording session for Verdi's Un Ballo in Maschera. I sung in the Junior Royal College of Music choir and on this occasion we were taken by coach to the recording studio and then waited for Pavarotti to arrive.

It was a fairly hot day in London. When he entered the studio, his presence I can only describe as being even larger than he was himself. He had his trademark handkerchief and he greeted the kids in the choir like long lost friends. I'll never forget the warmth and sincerity. His singing was unbelievable, I was standing a few rows behind him and I could hear him as if he was singing head on.

Thank you Luciano for an amazing career

JAVA stock ticker change - what's next?

Not since the proposed Java 2000 have I had a good whole hearted laugh at Java. My colleagues couldn't believe it either. When I was at Sun there were some interesting economic ideas about reducing costs and improving the stock price, many were based on financial statistics and few that I can remember had any impact. Maybe some new starbucks investors may accidently select the wrong stock?

So what is next, APPL to change to IPOD, MSFT to OFFICE, ORCL to DB. Does JAVA have to be quoted as JAVA(TM) on the ticker to keep trademark status? What happens if the coffee commodity industry hits a bad patch, would JAVA be affected?

Thanks Sun for some news that made me smile! I'll still be using Java for years to come.

Dirty secrets of computer engineering

Imagine taking your car into the shop, they call back saying you need to apply a manufacturers fix to the fuel injection system, it may not fail now but soon. Instead you say, "No way! I put one of those fixes on before and it took a week to get the car working again!"

OK, my example isn't perfect but in the software world we are trending towards the wrong end of trouble-free updates. Don't get me wrong, software updates in principal are great, I like useful new features, fixes to annoying problems. The problem is that each user now has more customizations, more add-ons and the applications are more complex and modular than before. The tested patch may never go through the number of permutations required out in the field

There is a whole industry of "Change management" like ITIL, to essential mitigate the risk of a change causing a problem. However I found very little about identifying the root cause and technology to solve it. If you have a configuration file, which by its very definition implies it will be changed, how do you manage updates to that changed file? What if your reseller also changed it? How do you diff an XML file? Does a registry or configuration database help or hinder updates?

My colleagues Nimish and Vinay collected some of the current approaches and a strategy for configuration updates. There is also a new open source web-based diff tool, JSBlend. If you have used graphical diff tools before the web based approach should look familiar. The industry still has a long way to go to clean up the current state, I just hope we are heading in the right direction.

Java on Fiesty Ubuntu - will anyone notice?

I've been busy working on our Web 2.0 release so didn't have time to update my laptop until now. I was generally happy with my Ubuntu breezy 64bit install, I had the JDK on there, Java worked in firefox 32bit, I could remotely display my screen to a projector and my broadcom wireless card even worked with ndiswrapper.

Now I had a few days to spare I decided to upgrade my system. I didn't get off to a good start going from breezy to dapper lost some of the packages I had. I then went from dapper to edgy but then my wireless would not work with ndiswrapper (kernel module had DMA allocation issues). Even the new bcm firmware cutter utility would freeze the OS and I had to remove the module from the module list.

I decided to continue with my final goal of getting to Fiesty. First it took all night, and then hung when the bcm cutter was installed. I repeated my previous removal of bcm43xx by booting into the live cd to get to my disk and then continued my update. ndiswrapper was not being loaded so with a quick modprobe ndiswrapper my wireless came to live!

Now for Java, I went to my firefox to http://java.com, the site appeared to be temporarily down. I then did java -version, hmm gcj not Sun Java. I then looked through my package maanger and found Sun Java. Both Java 5 and Java 6 could be downloaded. I went for Java 6

The install was straightfoward compared to the mess of faking ubuntu java dependencies before. I had to accept the DLJ license which was a little confusing as it referred to its own license version (1.1) instead of the Java I was using.

However, how many Ubuntu users will be actively looking for Sun Java. Maybe a clean install will ask but I think many developers will be oblivious to this new effort.

So now I have wireless and Java 6, If only there was now a 64bit browser plugin. I have been very tempted to cut my own, Juergen at Blackdown did a great effort but ran into many browser api compatibility problems. The painful part though is that the plugin is hardly a plugin at all, its essentially the JVM running embedded through some now very outdated APIs. While different groups did own the mozilla/netscape/java relationship before at Sun I think this got lost through many years of layoffs. Now maybe no-one owns it, or has little say to get the 64bit interface nice and clean :(

The unwritten story of open source java

What Sun finally did this week by releasing Java under GPL was a historic event. Using the GPL instead of yet another Sun license certainly makes adoption that much easier. But why did it take so long and why the change now?

I left Sun in 2004 after 8 fun years at Javasoft. Open Source Java was a annual discussion at Sun and in many conversations with open source advocates and companies. Many of the engineers I knew were for open sourcing Java as long as the motivation wasn't just to simply reduce headcount in engineering. However each proposal got shot down somewhere in the approval process. There were theories on how, who, why but ultimately the result was the same, at best a small compromise in the license but still no open source Java.

However competition is good, in the same way that C# ultimately made Sun continue to invest heavily in Java, the continued investment in projects like Harmony were always on the radar. In addition having a technical relationship with the IBM Java team meant that IBM used every opportunity to request Sun open source Java as every JSR lead was reminded as the defacto IBM comment in the JSR approval process was that IBM request a change in the license

Danese Cooper decided to work out the full timeline of open source Java The People Who Brought You FOSS Java as you can see its been a fairly long road and involved many people. Sun kept many communities involved including the Java champions which I have to give them full credit for, although it would have been even nicer if we could have dropped the NDA! So in closing, this is a great week, could have Java been open sourced earlier, yes but even late in Java's timeline it could be the real jolt that desktop Java needs. Do I think Java will fork?, no, maybe slightly, but nothing that would affect companies that standardize on a version anyhow.

Oh and Dalibor, I really owe you that beer now!

What PHP needs to learn from Java

We all know that there isn't a single language or platform that is totally secure, much in the same way that no matter how well you secure your house, its still possible to leave a door unlocked. However if your house has few locks, open windows and is in a bad neighborhood, shouldn't you do something about it?

The house I am referring to is PHP, great for prototyping and building applications quickly, but has a long way to go before it provides the automatic piece of mind Java does. We track vulnerabilities on all platforms and applications that use PHP are often the target of attacks. Was it the application developers fault, possibly, but there is little help for PHP developers to find out if they are really writing insecure code and without a security manager like Java, any small mistake can become a big exploit.

My colleague Ezra has started a new open source php security tool to audit php applications, phpsecaudit if you have any php code lying around check it out. We are looking for other contributors too. As for me, my first choice is still Java, even if it does take longer to create something the first time around.

Java and the Titanic

Nothing to do with the Mythbusters but the titanic really was outside Javaone today before being lifted on top of the Metreon.

Yesterday I spent several hours walking through the Java pavilion floor, lots of great stuff to see and some good giveaways by the vendors here. I also stumbled upon a photo opportunity of James with the Dukes Choice awards

Javaone: w/images Java SE and Java EE keynote pt2

Java EE 5 for many of you is a big step forward to unifying the Java EE platform. The removal of application specific deployment scriptors should be very welcome, something that has always been a barrier between moving between applications. Learning deployment in Geronimo for example is just one extra tasks developers don't want to learn. There was the obligatory Netbeans demo, although the demos also featured a text editor and command line deployment. As Ludo and Bill study below...

Bill quoted some numbers, that with all the Java EE 5 features the adventure builder demo is now 36% smaller (down from 67 classes to 43) but even more impressively a simple XML app was down from 5 files to 1. He cited the Glassfish project had reached 300K downloads.

The next section was the new features in Java SE. Its interesting to me to see the number of Sun Java EE engineers now working on Java SE features. Java SE was never got the respect it deserved inside Sun (imho), Java ME and Java EE due to their associated revenue got more of the attention. New Java SE engineers include Danny Coward who will now be the Dolphin Spec lead (a fellow brit) and Roberto for XML/AJAX.

So one of the more non-incremental features planned is a transition plan for Visual Basic developers to the Java platform, a project called Semplice. This is well overdue, especially given the Java-Sun settlement all those years ago.

Another feature was project Phobos, the ability to call Javascript to Java, following on from the petstore AJAX demo earlier in the day.

The rest of Java 7 may include native XML, friends, Swing application framework and Beans bindings, dyanamic language support. My first glance of native XML was that I didn't like it, it may be clever but including raw XML as per the slide make things more messy for me.

Finally if you don't like lines, Javaone still has teething issues with the new schedule tools. To get into the Java SE session took 10 minutes and then another 5-10 minutes to get out again!

Javaone: w/images Java SE and Java EE keynote pt1

Now listening to the Java SE and EE keynote. Graham is revisting the core features of Java 6. Which includes some GUI clean-up for windows and updates for Windows Vista.

The GUI looks nice and 10 years on the gray/grey rectangle is gone.

But as a Java Linux user I want a 64bit plugin, the number 4 feature on the web site. I doubt whether one will be around for beta2 in June for final release in October. Next up is Bill Shannon with Java EE 5 which was approved a couple of weeks back

Javaone news highlights. Open Source Java

So Jonathan has taken stage. The first item he wanted to discuss was Suns Niagara hardware sparc try and buy program so that you can now download free hardware.

Next came one of the conference sponsors, Ed Zander with Motorola who picked up a duke choice for a Motorola phone. Then the JDK java distro license change was introduced by Mark Shuttleworth of Ubuntu fame. Jonathan hinted about Ubuntu on 'servers', which of course ubuntu already runs on servers and then asked about support on ubuntu on Sparc. Mark did not confirm anything.

Next Marc Fleury came on with a red hat(beret) and hinted some collaboration with Sun and Redhat and that redhat was joining Netbeans. Redhat have been making their own eclipse edition so far.

Next Rich Green was invited to say that Sun is working out how to open source Java and not whether it would be open sourced. Given the time to open source solaris I wouldn't expect anything to happen for a while. I hope this is a real announcement and not playing for time.

Next Jeff Jackson is talking about Java EE 5. Which is already has an open source implementation and had the jcp members stand on stage and Jeet to discuss how to use blogs from the Glassfish community! ...

Next came J2EE and AJAX demos from Craig and Greg, knowing Craig the beer finder was a useful google mashup

Javaone news: Java on Linux the real story

One of the first Big community Javaone items this year wasn't the Open Source Java news but a new license for the JDK on Linux. Now I really believe this is a good thing, not enough, but an improvement. However, if you read Simon Phipps blog then Sun appears to be tackling something akin to the Berlin wall and funny enough re-writing history at the same time! Without giving away too many secrets here is the real story.

For reasons that date back to the 90s, and which were never meant to cause GNU/Linux a problem (as at the time it wasn't really on the radar), the Java platform has been licensed in such a way that GNU/Linux distributions couldn't carry it. In addition, the Sun-provided installer for GNU/Linux has, to be charitable, sucked.

Now for those with good memories, of course linux was around in the 90's, infact the original license for JDK 1.1 was friendly to many distributions and made many flowers bloom. Things changed with Java 2 and the SCSL license, of course every linux distribution wanted to have Sun Java on their CD and Sun had agreements with Redhat, Suse, Caldera and others. Any of the restrictive terms were due to Suns binary license and no-one else, for example the license click through on download and install was introduced later and enforced on blackdowns distribution. So Sun built the wall, and forced the click-through sucky installer and knew what it did.

"An unprecedented collection of Debian developers, Ubuntu developers, Sun engineers and Sun lawyers has spent months devising a new binary license for the Java platform"

I can't believe it took this much effort, kudos for those 'unprecedented collections' of Debian and Ubuntu developers for waiting, anyone who uses those distributions knows that they have invested effort into virtual packages anyway to handle offline distributions. Like Wei adds, does this mean fedora will be JDK ready, perhaps...if it does, it will be down to the effort of individuals of course, taking Java back to its roots.

Java: What's in a name

I read another blog entry complaining about the Java naming scheme. If that is the biggest issue with Java 5 today then I'm very happy. The complaint is that there are 3 names, Java 5, Java 1.5 and Tiger.

Well first let me tackle the Tiger bit. The Java codenames are all listed here. Tiger or 'Project Tiger' was the name used on a day to day basis, its a convenience instead of saying one point five point zero. The names are chosen as big animals (birds, mammals) for major releases (tiger, mustang) and little insects for maintenance releases. Graham Hamilton took this naming to heart and even launched a poll to decide on dolphin. The insects were all to be beneficial insects too, don't expect a Java roach to make an appearance any time soon.

The name mustang was initiated by a little in-joke by myself, both the product managers at the time drove a Ford Mustang. It wasn't hard to convince them that it was a good name.

The other numbering scheme was so nearly Java 2.0 for Java 1.2, Java 2000 was also on the table as it was the fad at the time. Who knows there may have been Java XP going down that road.

In all that time though the platform internal version has stayed consistent, we looked at how Solaris tackled this with SunOS versions and Solaris and decided the last people we wanted to upset were the developers who wrote code on the platform. Personally I think Java 5 as a name is at the right name at the right time.