(InfoWorld) - One of the survivors of the early days of Linux has changed its name and folded its software division in favor of doing business entirely on the Web.

VA Software said Thursday it has changed its name to Sourceforge Inc. and will focus the company exclusively on its Web properties, such as the SourceForge.net site that hosts open-source projects and the Slashdot news aggregation site, which is popular with technology enthusiasts.

The company said the moves are in response to the dynamics of Web 2.0, the new business model in which traditional software is beginning to give way to advertising-funded Web-based services.

VA Software last month quietly sold its enterprise software business to CollabNet, a company that offers development software for geographically distributed teams, in consideration for an equity ownership in that company. The newly branded Sourceforge will make its revenue from advertising sales and product sales from its geek merchandise e-commerce site, ThinkGeek.

Other properties that are part of Sourceforge include another open-source project site, Freshmeat, and enterprise Linux resource site Linux.com.

VA Software began its life as VA Research in 1993 and later (and more notoriously) became VA Linux Systems, one of the early distributors of Linux whose stock jumped to $299 on the day of its December, 1999 IPO. But its Linux business soon flagged, and the company had more success in software development and acquiring the Web properties that still operate today.

SourceForge.net in particular is a successful site for the company and one of company's the most recognizable brand names. The site hosts more than 100,000 open-source projects and has more than 1 million registered users.

Sourceforge also reported is third quarter fiscal 2007 financial results Thursday. Revenue was $10.3 million for its third quarter of fiscal 2007, a 30 percent increase from the $7.9 million the company reported for the same period last year.

The company reported net income of $6.5 million mostly from the sale of its software business. After years of not turning a profit, VA Software recorded its first quarter of profitability in its second fiscal quarter of 2006, which ended Jan. 31, 2007.

ADVERTISEMENT

IBM Information On Demand 2006
Industrial Industry Leaders, please join us at IBM's premier information management global event, IBM Information On Demand 2006, October 15-20, Anaheim, CA. More IBM business and technical solutions content in one place than ever before! Select from over 800 sessions. Register today!

Google News and Techmeme are two services that I use extensively every day. They are also two services whose technology is way over my head and are both eminently impressive. But Google News could, in my opinion, learn a lot from Techmeme, which has no rival when it comes to tech buzz aggregation. And Techmeme could learn a few things from Google, too.

It should probably be noted that Google News and Techmeme have very different aims. Google News aggregates news from across a broad spectrum of categories, mostly from mainstream sources. Techmeme, on the other hand, highlights buzzworthy news from a focused niche (technology), mostly from blogs. But they are very similar beasts. They both aggregate news very well and algorithmically decide what are the top breaking stories of the day.

Display

The most immediately noticeable difference between the two services is in their display. This is one area I think Techmeme really shines. Techmeme's display is elegantly simple, easy to read, and usually compact (except when a monster meme evolves, such as the iPhone announcement). Google's display is also compact an easy to follow, and their sheer number of sources likely makes it impossible to list every contributing source at once the way Techmeme does.

But Google's display does some things I don't like, as well:

1. It constantly shifts - This again may be a product of the massive number of sources, but Google News can never decide who controls the meme. I am guessing that Google's algorithms rely more on semantic analysis and Techmeme's rely more on linking, which probably makes it harder to decide which story to keep on top, but every few minutes, Google news looks very different, which can make it hard to find the story you were just looking at. Techmeme, on the other hand, generally decides on a few top memes over the course of a day, and feeds them (this process is, of course, exaggerated since many -- most? -- tech bloggers watch Techmeme and chime in with their own two cents on popular stories). It shifts, but not much and not every few minutes.
2. About those sources... - Google News gets content mainly from mainstream newspapers, which means a ton of duplicate sources from the Associated Press, Reuters, and other wire services. When you see Google News quoting 3,500 news articles in a single meme, how many of them are repeats? Listing all the duplicate sources certainly has its place (sometimes a wire article will be presented with an optional trim -- meaning you can find a more detailed version elsewhere, or sometimes you just want to find the article on a more attractive page), but I wish it was divorced from the main service. Techmeme, because it handles mostly blog content, rarely displays duplicate articles. It's nice to be able to scan through multiple viewpoints from different sources on the same topic.
3. Limited number of stories - Google News shows a maximum of nine stories per section. This makes sense for a generic page, where multiple topics need to be displayed on the page at once. But the restriction makes less sense for a customized page that could theoretically only display one or two topic areas. Techmeme, because it is more narrowly focused, has the luxury of displaying many more stories at once.
4. River of News - Techmeme has a River, which shows breaking news stories chronologically throughout the day. This is an incredible feature that allows you to keep on top of buzz as it heats up. As I said, it can be hard to keep on top of news with Google's service since it shifts so often. A timeline of the day's news would be a great addition.

Timeliness

Techmeme feels more timely than Google. Even though I said Techmeme keeps its top stories in place longer than does Google News, it also seems to find new memes within 15-30 minutes of when they break. Google News finds stuff quickly, but often displays articles that are 18-20 hours old because the meme is still being pumped (many times by duplicates from the wires being pushed out to newspapers late).

Archives

The constant shifting of Google News' homepage and the limited number of stories you can display at once per category often means news slips by without being seen. The lack of a proper archive means some of those stories may never be seen. If I want to know what yesterday's headlines were -- or even this mornings -- I am out of luck. Techmeme, provides a very comprehensive archive. Enter any date and time and instantly view how the Techmeme page looked at that moment in history. What Google News does offer, is archive search...

Search

My biggest gripe with Techmeme has always been its lack of a search feature. In order to find the link to the iPhone announcement meme I linked to above, I had to first remember that the Steve Job's MacWorld keynote was on January 9th, and then load up that page at the correct time of day. It would have been great if I could have looked up "iphone announcement" and been directed to the blog posts I was looking for (we'll pretend I was actually after the posts and not the meme link).

Google News has always had search, and great search because this is, after all, Google. The only check against it was that it didn't go back very far in time. You couldn't search for last year's news, or in some cases, last month's news. But Google recently introduced News Archive Search, which allows users to search an archive of news stories back to the 1920s (or before). The results are a sometimes a bit odd and out of order, but it is infinitely useful as a research tool regardless. The search also has a timeline feature, that attempts to assemble a timeline of the top news throughout history on a certain topic. Sometimes its choices are rather perplexing, however. Check out this timeline of Einstein and you'll see what I mean.

Conclusion

These are both services that I use every day and rank among the most useful in my arsenal of online tools. Neither is perfect, but both are marvels of engineering (to me, anyway) and the people behind them deserve my thanks for creating something so integral to my daily routine. I think the two services can learn from other another, though, and each become even better at what they do.

After over 6 months of hard work the MbUnit 2.4 went to RTM tonight. Between the RC and the RTM.

Work items

Bug

• [MBUNIT-27] - ExpectedExceptionAttribute does not verify message
• [MBUNIT-79] - "Console.Out" tab not cleared on assembly reload
• [MBUNIT-103] - TD.Net doesn't report mismatched version
• [MBUNIT-104] - DataFixture ForEachTest XPath fails if it has a /
• [MBUNIT-105] - HTML Report encoding sucks
• [MBUNIT-107] - Assert.AreNotEqual() does not increment assertion count
• [MBUNIT-109] - AssemblyCleanup requires a static class
• [MBUNIT-112] - Failed Combinatorial Tests do not call tear down
• [MBUNIT-117] - Double-clicking .mbunit project file throws exception while loading assemblies into GUI
• [MBUNIT-121] - Error when report type is supplied but no report folder

Improvement

• [MBUNIT-11] - FixtureSetUp Log missing in GUI
• [MBUNIT-77] - Tree order is not sorted
• [MBUNIT-122] - Console help should only be shown when necessary
• [MBUNIT-124] - Run button should not be enabled if no assemblies are loaded

New Feature

• [MBUNIT-63] - RowTest and decimal parameters
• [MBUNIT-65] - Open MbUnit to Last Test
• [MBUNIT-118] - Extract Embedded Resource Attribute
• [MBUNIT-119] - Test non-public members of class. Also Test non-public classes.
• [MBUNIT-120] - Use DateTime in RowTests
• [MBUNIT-123] - Specify path for AutoRunner reports

My thanks to the folks that worked hard on this release, Graham Hay, Vadim Kreynin, Julian Hidalgo and Phil Haack.

In order for MbUnit to reach 2.4 the thanks once again to all the people that worked hard, Graham Hay, Vadim Kreynin, Julian Hidalgo, Phil Haack, Joey Calisay,  Ben Hall, Jeff Brown, Tomáš Chaloupka, Albert Weinert, Marc Stober, Johan Appelgren, Cathal Connolly and Todd Menier.

Highlights

• .NET 1.0 support dropped.
• .NET 1.1 support in 2.4 merged into the main line (No need for MbUnit.Framework.1.1.dll)
• Added greater support for .NET 2.0 including (MbUnit.Framework.2.0.dll)
• Database rollback (now no longer need COM+ if your using .NET 2.0).
• Private method testing (see Vadims articles here and here).
• Generics in certain Asserts.
• Support for NUnit style explicit attributes.

 As well as a great deal of bug fixes and functional improvements.

The road ahead.

2.4 signals the end of major releases for 2.*, interim releases with tweaked functionality and bug fixes will still occur to 2.4 (starting with 2.4.1) but the next major release of MbUnit will be v3 codenamed MbUnit Gallio. In the coming weeks the core development team will be stepping up through the gears on MbUnit Gallio.

Last week Ernie Boothannounced a SilverlightBrowser plugin for Reflector. The plugin allows you to browse web pages that contain Silverlight elements and display their constituent Javascript, XAML and assembly files.

Ernie released the 0.1 Alpha code so people were free to modify it. The way the 0.1 code obtained the .js, .xaml and .dll files was by searching the HTML and Javascript source for specific patterns. For example it would search the HTML file for 'Silverlight.createObject' and extract the .xaml link from its parameters. This is similar to how Lutz's SilverlightLoader plugin works.

The problem with this technique is that it assumes that the 'Silverlight.js' and API (which comes with the Silverlight SDK) is being used. Unfortunately this won't always be the case. Some people may include the <object> or <embed> tags directly in the HTML or embed the XAML in a <script> block. A good example of this is Sam Ruby's Inline SVG in MSIEdemo. This demo creates a Silverlight <embed> element and converts the SVG content into XAML on the fly!

I've modified Ernie's code so that is searches the WebBrowser object model for the .js and .xaml files. This means it is no longer making the assumption that 'Silverlight.js' is being used. It doesn't yet support XAML that is emended in a <script> block. I'm hoping to have it working with Sam's demo in the next version.

I have kicked off a community project for the SilverlightBrowser plugin on CodePlex (with Ernie's blessing). You can download latest version of the plugin from here.