The WSRP 2.0 spec is also expected to finalize this year.
With both key specifications in portal land undergoing a major revision this year, we bound to see a lot of action. So stay tuned !
]]>With a dedicated, global team of engineers working round the clock backed by a excellent management team, I believe we have delivered a product that can give our competition a run for their money :)
Portal 7 places a heavy emphasis on today's collaborative or Web 2.0 technologies- Online communities (aka social networks), wikis, blogs, file sharing, syndication, ajax.
We have also worked on many not-so-cool-yet-crucial features - Administration, Management, CMS integration and Enterprise Application Integration. For more info, check out the portal server product page
I have also started a new blog called The Portal Zone specifically focused on the Sun Java System Portal Server. If you are a developer using the Sun Portal or in the process of exploring the product, you ought to find this blog useful.
]]>I wonder if the same thing is possible with software. Is is possible to develop software using an anarchist approach ? Can a project open up cvs commit access to anyone on the Internet and expect a coherent piece of software to evolve over time?
I wonder.
]]>I see a few hundred geeks waiting patiently in front of the registration counter with T-shirts proudly displaying their corporate affiliations. I feel bad. I didn't wear anything to display mine.
10:00 am
The inaugural session is delivered by Atul Chitnis who I suspect is one of the pioneers of the event. The event is kicked off in the
traditional Indian custom of lighting a lamp, the lamp being lit by five LUG leaders from various parts of the country. Atul remarks
something about the lamp representing knowledge and that it takes more than one person to spread it. Smart !
10:30 am
Keynote by the Kernel guru Alan Cox begins. Contrary to my fears, the session did
not dive into dissecting kernel modules. Instead, it was a fairly high level introduction to open source and how to get started with OS
contributions.
11:30 am
I walk into the expo area to check out the stalls. Intel is a diamond sponsor and I am wondering what Intel has to do with open source
software. Instead of a stall, I just find a large Intel logo displayed and some chairs for conference attendees to sit and relax ! So much for Intel
and software. I share a joke about it with a fellow Sun employee and move over to the other stalls.
11:40 am
I spot the colorful Yahoo stall. I don't find much information in their displays and posters, so I ask the guy what yahoo's contribution to
open source is. He tells me that they use a lot of open source software internally. I guess that's valid enough. Using open
source software is also a means of contributing to the community in some sense.
Yet one expects much more from a company like yahoo. So I rephrase and ask if yahoo contributes any code to any open source community. He says they are considering contributing code to some open source tool (I forget the name) in the future. Then he asks me what Sun's contributions are to open source. I grab the opportunity to explain about opensolaris, netbeans and glassfish and invite him to visit the Sun stalls across the hall.
11:45 am
I am at the "Free and Open Source Opportunities in India" session by Danese Cooper. She surprises the audience by wearing a traditional
Indian salwar. She urges developers to contribute for open source. She also talks about
the abundance of opportunities in India. She educates the audience about how Intel offers the "Platform of choice" to developers. If developing
nations want to deploy FOSS solutions Intel is ready to support it,she says. Fair enough. But I still don't understand what Intel is doing at this
conference.
12:15 pm
Venkatesh Hariharan aka Venky's talk on "The Political, Cultural and Economic Relevance of Open Source" is refreshing. He talks about
some very interesting aspects of Open Source and how it's relevant to the Indian economy. I'll probably need a whole blog entry for it.
This is probably the best session that I attended.
2:30 pm
"Google and Open Source" by Zaheda Bhorat begins. The whole Google thing looks like a recruitment and PR drive. They have a "We are hiring"
caption right in their presentation slide. Google's contribution to open source is that they run on Linux. I wonder if that's Google's contribution to
Linux or Linux's contribution to Google. And of course it seems they have open source examples of how to use the Google API.
The next 30 minutes is about the Google Summer of Code. Zaheda discusses how the whole stuff works and how successful the program was. She then gets a student on stage who shares his experiences being a participant in the Summer of Code. I hear him say "Thank you google" at least thrice.
3:30 pm
I got to leave to office. I have promised my manager to catch up with some work. I have anyway had too much for the day.
Goodbye FOSS.in. See ya tomorrow.
Web continuations can change the way we think about web applications. In the servlet world, we develop web applications with a callback paradigm. i.e when a user clicks a link, the servlet container calls the doGet() method of our servlet. Web Continuations invert control (The notion of control is in the mind of course ;). This means our server side code would call the container to dispatch a page to the user and wait till the web container gets back to it with a result. When the container gets back with the result (which means that user has initiated one more request) , the code continues execute from where it last left.
Web continuations makes us think in a linear fashion i.e we can think of HTTP requests and responses as printfs and scanfs over the network and web applications can be coded like command line applications. It also makes web applications inherently stateful. Since our code continues exection from the last line it executed before dispatching the page, we don't have to do any additional steps to store the state(no storing/retrieveing data from HttpSession)
While continuations simplify web applications I am not sure if they would remain simple if the web app deals with complex navigations. Moreover I think most developers have stopped coding servlets. We use either a tool or a framework and deal with web applications at a higher level. And lastly frameworks like RIFE have proved that web continuations are possible in Java too. So even if you do want to use a continuation server, you don't need to think beyond java.
]]>The central theme of the book is "productivity". Bruce argues that Java has become too complex and cumbersome to achieve simple tasks like writing web-based CRUD applications (web-apps that typically front-end a database with Create/Retrieve/Update/Delete operations). He predicts that parts of Java, especially those that deal with web application development will be replaced by a dynamic, productive framework like Ruby On Rails(RoR). The book also covers the Ruby language pretty well with a comparison between Ruby and Java.
Well ...to be honest, I would say that Java application development has never been as productive as it is today. Yes , seriously .. and I think a lot of people would agree with me on that. Long before joining Sun I used to work for a consulting company. And I don't remember any of our customers choosing java because it was the most productive option around.
When I was working on my first server side Java project, my friend used to work on this cool technology called the Allaire Cold Fusion Studio. I would write database connections pools and templates (as JSPs still in 0.9 we used our homegrown template engine !) and he would put a few CFML tags into a web page and achieve the same. The productivity of Cold Fusion was amazing. In fact,at that time, my company had more people working on Cold Fusion than Java.
Yet, one of our biggest customers who came by opted for a Java based solution. Why? Vendor independence. They were attracted by the fact that they didn't have to be tied to a particular vendor for their solution.
And then there was this time when I had to code a little GUI tool in Java. My VB/Delphi friends laughed when they saw me hand-coding my GridBagConstraints. The truth was there was no Java IDE available to actually generate a decent responsive GUI ! Yet the application was coded in Java because the tool needed to be platform independent.
The point is that developer productivity, though important, is not the only reason for choosing a platform. If you want the least lines of code, Perl would be more productive than Java. VB might be the fastest tool to develop a complex UI. RoR might be the fastest way to create a CRUD based application. And when you compare a statically typed language like Java with a dynamically typed language like Ruby, you will always find the latter to be more powerful.
On the other hand, times have changed for the Java developers. Today there are *plenty* of good tools out there for us to use. We have stopped handcoding our GUIs. We can drag and drop components into web pages. We can right click and "convert anonymous class to inner class". Today we can easily do things that we have struggled to do before. If not anything else, we have definitely become more productive.
Speaking of tools, Bruce does not seem to be a great fan of tools or generated code (And that's where I think he misses the point). He states that even if you are using a tool, you need to understand the code that your tool generates and that generated code is difficult to maintain. I don't agree with that observation.
If a tool is good enough,one does not need to understand generated code(Ask a VB developer). And generated code can actually be easier to maintain. How many of us take a look into RMI and JAX-RPC stubs that our tools generate ? Even JSPs are generated code, but I don't remember ever trying to understand it.JSF takes the paradigm a step further.
So instead of developers moving away from Java what we are likely to see is that our tools get better and better to make us more productive. Yes, We might never be able to reach the productivity level of a dynamic language/framework like Ruby. But that difference would be more than offset by the inherent advantages that Java provides and the huge amount of tools/frameworks/libraries available.
]]>You might find it useful too. You can check it out here: Quick' n Easy Portlets using the Sun Java Studio Creator 2( I used wink to create the demo. It's an awesome tool !)
Portlets built using Creator 2 are JSF based. Creator 2 also bundles Pluto, the JSR 168 reference implementation. Pluto is used as the runtime to test the portlets. Creator 2 certainly looks promising for portlet development. You can download Creator currently in Early Access from here: Sun Java Studio Creator
Recently Howard M. Lewis Ship, the creator of Jakarta Tapestry, brought to my attention that Tapestry 4.0 has incorporated full native support for JSR 168 portlets. Have not tried it out myself, but from the docs and some tidbits I got from Howard, looks like a neat implementation. Tapestry 4.0 is currently in a late beta stage. You can get it from the Tapestry site
]]>“Using the XMLHttpRequest Object and AJAX to Spy On You” is plain FUD. Ajax has neither increased nor decreased privacy concerns. In fact Ajax is just a clever way of combining already existing technologies to create better websites. The author's claims seem baseless.
Someone pointed me to this “Elevator hack”. Apparently, certain models of Otis elevators can be made to bypass all floors except the one you want to go to, by simultaneously pressing the “door close” and the floor number buttons. Hmm .. now that's something worth trying :)
I have heard other claims too; Some bizarre, like jumping up and down inside a moving elevator. The elevator, it is said, has sensors to detect abrupt jerks. If you can create sufficient impact by jumping, you can fool the sensor to bring the elevator to a screeching halt. The only elevator hack (feature!) that I have seen working is pressing the door close button continuously to bypass desired floors. However, I have seen only an authorized operator doing it and I didn't notice if he had a key.
While most of these so-called "elevator hacks" seem to be urban legends at best, the next time you find yourself alone in an elevator you know what to do ;) ]]>
So how do we ensure that today's systems can be best utilized tomorrow ?
We can selectively use parts of the system in the future.
We ensure that we can always “talk” to these reusable components.
That's (my take on) SOA. It's not Webservices, It's not WSDL, It's not XML. Heck, It's not even HTTP.
SOA relies on principles like clearly defined interfaces, well-defined entities, independent responsibilities etc etc... which resembles guess what ? Yeah. Good old OOP.Frankly, the more I think about SOA, the more I find it similar to OOP.
As for Web services, web services can be used to create SOA as much as Java can be used to create an Object Oriented Program.SOA and OOP are principles. WebServices and Java can be used to implement those principles.To say SOA means Webservices is as naïve as saying OOP means Java.
Hmm ... am curious...are there any notable REST based implementations of SOA?
]]>Why do I feel so ? Let's try comparing the formal web service approach to the poor man's approach:
To communicate with a Web Service you need to know message formats.
Write a WSDL for every web service that you need to expose. The WSDL would define the message formats, datatypes, transport protocols, and transport serialization formats that should be used.
There is a well know message format that everyone needs to use. And THIS is the format.
Finding the end point to connect to
Discovery can be registry, index or peer-to-peer. UDDI is a standard registry based approach. UDDI exposes Technical Models or tModels. Use UDDI APIs to query the registry and obtain WSDL of the webservice that you intend to connect.
Scrounge through the web site and search for an orange button labeled "RSS" or "XML". When you find it,Click on it. If you want to do a wild search, use a feed search engine like Feedster.
Authentication, Trust and Security policies
Use Web Services Security with a Token Profile.
Security ... huh ?
.Connecting to the service.
Write code that can send SOAP messages in compliance with the WSDL, after you have obtained the WSDL by making UDDI inquiry calls to the UDDI registry. Make sure to satisfy all the security mechanisms.
Open your favorite feed reader, Click on the big "New Feed" button, enter the URL and click Finish. Ok, If you want to code it yourself, write a java program (or even a shell script ) that can get the XML from the endpoint and parse it before displaying.
There you go !. As you can see, RSS follows the web services template but with a highly simplified approach. IMHO Simplicity is the essence of RSS's success.
Now, don't get me wrong. I am a great fan of Web Services as well. I have written a lot of web services code. I have even dared to read a few documents from W3C :). Web Services are complex because they try to represent real world business processes, which are inherently complex.
However, my opinion is that real world adoption of web services will occur by radically simplifying the web services model ...like RSS. Simplified models would further evolve to more complex ones, taking each need into consideration( compare RSS 2.0 and 0.9)
Hmm .. hey wait a minute. There was something I was wanted to announce through this blog entry and that's the reason I am writing this. Oh yeah ! The Portlet community has just setup up an RSS feed. You can find it here.
Check it out and enjoy Webservices...the poor man's way ;)
]]>