The JDK will be available for download from openjdk.dev.java.net, and (as of today) the HotSpot Virtual Machine and javac are available. The rest of the JDK will posted there in early 2007.
]]> Sun had announced their decision to do earlier in year at JavaOne. This was the first JavaOne without Scott McNealy as the CEO, and methinks Sun finally realized what IBM did years ago about how to make money from software. You can't make money- at least big money- with software infrastructure products and tools.]]>The one I like the best? "Guns don't kill people, the SOA WS-* stack kills people."
]]>Besides the demos, there were visits on stage from Mark Fleury (JBoss, soon to be Redhat), Mark Shuttleworth (Ubuntu CEO, and part time cosmonut), and others whose names I can't recollect right now.
Stay tuned for more reports!
]]>I've had some pull parser experience - primarily with kXML, so it wasn't a new area. However the StaX API has some neat ideas in it which I liked, such as filters. StaX allows you to define filters at the parser level, such as a filter for specific namespace. Using filters, the parser can skip over the stuff in the XML that the application doesn't want to look at... thus leading to far cleaner, and more performant code. Jeff's StaX presentation is on his blog.
]]>Step 2. Write mocks for the components.
Step 3. Write the glue code that ties the components together and build your complete application. At the end of this process, you should have a fully functioning system. True, one that does nothing useful in particular, but one that works end to end.
Step 4. Write tests- unit as well as feature - that exercise this system.
Step 5. This would also be a good time to add a (continuous) build framework, say built using Ant/Maven and CruiseControl that fires off nightly builds and tests.
Step 6. Now hand off these components to individual developers for the actual development.
At any point in time, the whole system should build *and* run. The tests (defined earlier, and refined with time) act as regression. At the end of the process, there should be far less integration issues, and faults would be easy to isolate to individual components by replacing a component by its mock (say using a factory, or a Avalon type component framework)
This approach is similar to a children's coloring book in which the outline of the picture is drawn first, and then the child (or different children) can color in each part independently- hence the name.
]]>It wasn't until Google came in that maps become sexy again. Google has been able to take something that has been around for a while, redo it in a much better way, and create excitement around it. It has done it with search, with email, and now maps.
Other than a really well done interface, Google was able to get developers excited about maps to such an extent that it spawned the unfortunately termed, but incredibly exciting 'mashups' (see Chicago Crime and Free Wifi for examples). The earlier mashups reverse engineered Google Maps, though now Google has released a documented API.
Yahoo too has released its API, and its own catchup mashups (For eg, SF Dog parks).
This kind of stuff is the future of Web services, but more on that later.
Google then upped the arms race by first adding a satellite view of maps, and then a neat 'hybrid' view that combines both. Finally it added Google earth and a limited Google moon. It was now Microsoft's turn for catch up with Virtual earth
And now A9 maps has the latest new 'feature'- street level maps of major cities, and has a fleet of trucks taking pictures of the rest of the country… ambitious to say the least. I remember reading about Google experimenting with similar technologies- I'd guess we will see something similar from Google sometime soon.
What's at stake here other than who has the coolest technologies? Geographical searches and map visualization has a lot of interesting applications, and many with a lot of commercial value. The first and obvious one seems be to local search + yellow pages.
]]>There are a range of techniques that can prove useful here:
]]>In this introduction, I mention two application area where Web service technologies are, and will be very significant. However, increasing it now seems that it might not be the same set of technologies.
]]> The two areas are:The first refers to all the EAI type stuff Web services does, or is capable of. This is also where there is a lot of vendor activity, and where all the WS-* specifications are targetted.
In the second area, REST is becoming increasingly significant. It is getting more grassroots support, but a lot less vendor support. In an earlier post, I mentioned how Yahoo's new API is REST only, and that Amazon has seen most of its developers prefer REST to SOAP. REST is much easier to use, and would even scale/perform better than its SOAP counterpart.
]]>REST (Representational State Transfer) is quite simple and elegant, with its resource 'nouns' (represented by a URL) and the four HTTP 'verbs' - GET/POST/PUT/DELETE mapping nicely to CURD operations. However while I can see it being successful in applications like the Yahoo search API, it is, IMO, not very appropriate when used in the context of integrating more complex applications.
]]>One of the project chapters is about a personalized portal that integrates content from other websites and services using RSS and Web services. The chapter introduces good design practices while integrating Web services with your web applications, and where they fit in an typical MVC architecture (Web service invocations update the model etc.), but more on that in a later post.
]]> Update [March 04, 2005]: A sample chapter from the book is available here, and the the TOC here ]]>The Google Scholar FAQ has more information on this.
]]>I just installed it, and have yet to play around with it. Some things I'm curious about, and some disturbing (?) questions it raises:
]]>Google, I'm sure, has a lot more ambitious plans for this. The name (GoogleDesktop, and not something like DesktopSearch), implies that they have bigger ambitions than just being a better way to do search on the desktop.
PS: Slashdot has a discussion on it too today
[Update 12:23 PM]: After Google Desktop had indexed some emails, I went ahead and removed them (permanently) from Outlook. On doing a search again (after a few minutes), I found them in the search cache- complete emails and all. Disturbing. When does this cache expire?
[Update 1:41 PM]: Guess I didn't RTF-FAQ... From the Desktop Privacy FAQ:
Q. How do I prevent items from appearing in Google Desktop Search results?
A. If you want to prevent something from being found by Google Desktop Search, you can do any or all of the following:
Note that deleting an email from your email client does NOT delete its copy from Google Desktop Search. Nor does deleting a document or spreadsheet remove the stored copy of that document or spreadsheet that Google Desktop Search has created. This enables you to retrieve copies of things that you may have accidentally changed or deleted. If you want to delete the Google Desktop Search copy as well, you have to do so from within Google Desktop Search.
Oops. My bad.
[Update Oct 15, 3:04 PM]: A discussion on slashdot today talks about the issue I raised yesterday, namely being able to snoop on other users on the PC. Not an issue (or shouldnt be) if its a home PC, but could be a problem for any other situations. You can offcourse, configure it to behave in a sane way, like not index pages from specific websites, or folders, or not index any web pages at all. Besides, it only indexes content that is accesible normally, though it does make finding such stuff much easier. I had fun today with running it with random search strings, and seeing what it brought up.
Kind of like finding things under your couch cushion that you had forgotton about. Some loose change..and some icky stuff :-)
And I got it today.
]]> That's it. Nothing about warranty periods, how did it break, or the usual crap you get from bigger companies. Sony "support" is a case in point. I had a problem with the wireless card of my Sony laptop, and I was bounced around by one clueless service rep after another, made five phone calls to Sony support, and the problem was still not fixed. A couple even told me to buy a Sony brand wireless router fix the problem! There was no talk about sending a replacement wireless card, and all their system was designed for was to figure out if I qualified for the warranty, and if so, how they could wiggle out of it. Mind you, Sony makes good products, and if not for that, I'd never buy another Sony product.Innopocket? I'll buy anything from them again. Anytime.
]]>I kept hoping that this one won't catch on, but no such luck! Vendors have rushed to clasp yet another acronym to their bosom, and RougeWave even announced how their LEIF framework participates in the SODA development process.
Frankly, this leaves me in two minds- I agree that formalizations in design and development processes is important; but the rate at which the acronym/specification-churn happens in the Web services/SOA world leaves me (and I'm sure a lot of others), with a feeling that a lot of stuff here is over-engineered and over-hyped. This, to no small extent, is why even after four years of Web services, we still have a lot of programmers very leery of it.
References:
[1] Gartner. Service-Oriented Development of Applications: SODA Pops. http://www.gartner.com/webletter/bowstreet/art5/art5.html
]]>Oh, and by the way, you can submit feedback to GMail here.
[Update: Aug 30, 2004] Seems you can move read email out of the way... by archiving it. The archived email shows up in searches and in the 'All mail' option, but not in the Inbox. Thanks a lot for the tip Kevin Dangoor!
]]>