Hello, Marc Clifton writes:
I am releasing a preview of MyXaml 2.0 so people can get a sneak peek at what is coming down the road. After evaluating the MyXaml 1.0, taking into consideration customer feedback, and looking at how to position MyXaml 2.0 for future endeavors, I have reworked the code base with several architectural and implementation changes. Please note, this is a preview version. Do not replace MyXaml 1.0 with this preview version, as there are several features of MyXaml 1.0 that have not yet been migrated over to 2.0. Creation Of A Core Parser Several of my customers have requested that MyXaml not be coupled to the System.Windows.Forms namespace. Separating the parser from the presentation layer makes a lot of sense. With 2.0, you can use the parser in contexts that do not require a presentation layer. You can also apply the parser to different presentation implementations-web and compact framework, for example. Plug-In Extenders The System.Windows.Forms namespace must of course still be supported. Issues such as SuspendLayout and ResumeLayout, working with data binding, image lists, etc., will be implemented using the new extender feature. I haven't yet migrated all of the .NET 1.1 ""helpers" from MyXaml 1.0 to 2.0, but you can see an example of supporting the Suspend/ResumeLayout method calls for SWF by using an extender. The extender feature will also be used to manage any implementation differences/improvements found in 2.0 and support for inline code. The extender feature is also going to be relied on to help work with third party toolkits that aren't declarative-friendly. By using a combination of extender and component abstraction, it should be possible to overcome all of the problems present in these third party toolkits. The idea is to then offer different extenders for different toolkits. For example, there are five or six flavors of docking managers out there. The goal of the plug-in extender framework is to be able to use the same markup source, plugging in the specific docking manager extender for the third party toolkit you're using in the application. This helps to reduce "vendor lock" and even more importantly, problems that arise when a vendor releases a new version that breaks existing implementation. The declarative aspects of your application should be immune to these breaking changes. Instead, only an update to the extender plug-in needs to be made. Exceptions MyXaml 2.0, rather than using Trace statements to indicate errors, now throws exceptions. In the preview release, there are 27 different exceptions that the parser generates. While I'm not particularly fond of exceptions, many of customers (and admittedly I too) have wasted a lot of time tracking down object graph problems which is only noted by a trace message buried deep in the instantiation process. Unit Testing Have 27 different exceptions doesn't help if you don't test those exceptions. MyXaml 2.0 includes over 90 unit tests. These tests validate the parser's functionality, exception generation, and provides simple but useful examples of the features and object graphs that can be constructed. The unit tests require the Advanced Unit Test (AUT) tool, one of my other open source projects. Code Path Coverage The main property setter method presently contains 56 different valid code paths (by valid, I mean code paths that don't result in an exception). In addition to the unit tests, there is a test fixture dedicated to ensuring that each of these code paths is exercised (this is another reason for the AUT requirement as the test tool). Documentation I must admit a certain frustration with other open source projects and the lack of code documentation. MyXaml is small but complicated, and I have gone through the code with a fine toothed comb and documented every field, property, class, and method in this release. The result is a very nice nDoc generated CHM file. This file and the code documentation should be of great value to people who need to understand the working of the parser, especially when a third party assembly is causing difficulties in a declarative context. Old Features, New Features MyXaml 1.0 supported the compound property syntax found in Microsoft's XAML. This support has been eliminated. As to new features, this release includes support for nested class instantiation (see the unit tests) and assignment to properties whose type is Type--a useful feature. Licensing Customers that have purchased a MyXaml 1.0 license can upgrade to MyXaml 2.0 without additional licensing. Feedback Your feedback is desired! I'd like to see markup examples that should work but don't--either they throw an exception or they generate unexpected results. Also, let me know what you think of the extender architecture--it's very simple. Is this a good approach? If anyone is interested in developing an extender for a particular third party toolkit (docking managers come to mind), let me know! Conclusion As you can see, a lot of thought and care is being put into this release. Many people ask, "why use MyXaml instead of XAML?" I think you can see from the above discussion that declarative programming isn't just about vector graphics and object graphs. It requires an architecture that supports what people are trying to do declaratively--work in different environments, offer customization for application specific needs, and address real world issues such as complex third party toolkits, all in a well documented and rigorously tested open source framework. That's the MyXaml difference. Download the preview version of MyXaml 2.0 here. Source: http://myxaml.com/marcclifton/archive/2005/03/11/1717.aspx - Gerald ------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click _______________________________________________ xul-announce mailing list xul-announce@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xul-announce
