Opera Dragonfly

The weekly snapshot of Opera Kestrel included a few fixes to bugs that caused issues in Opera Dragonfly. The most prominent one of these is that persistent cache should now work in Opera Dragonfly, enabling offline mode to work as it was designed. This should make Opera Dragonfly much more useful.

Another bug that has been fixed, and should be coming soon, is that Opera Dragonfly will be useable even when JavaScript is turned off. Once this fix lands, it will be possible to debug how your web page or application works without JavaScript

In further news, we are doing our best to try to squeeze the single windowed docked mode into Opera Dragonfly alpha 2. This hopefully wont delay what we've already planned for the second alpha. It wont be the final solution, as we need to do some interaction design work on how it will work differently to the separate window solution. As there is less space to work with, it will likely be more optimal to have a different view. The initial work will likely just make it work in that view, and test out the functionality given to use from the Desktop team.

In a post alpha 2 release, we hope to redesign and optimise the UI somewhat, and work on keyboard accessibility. The ground work for the later has already began. I've been reading up on WAI-ARIA, and it looks like something we can put to good use. As well as making controls accessible to assistive technology, it should make our keyboard navigation work more like a native application for those controls. There are a number of roles for components that stand out instantly as useful for Opera Dragonfly, including tree, and treeitem for the DOM source code tree, toolbar, button, search and perhaps checkbox for the toolbars, tabpanel for the tabs, breadcrumbs for the DOM path and so on. It looks like ARIA should be something that isn't too difficult to learn or apply.

One of the great things about developing Opera Dragonfly in Web technologies (except being a useful exercise in finding Opera bugs), is that we can design only for the Web standards provided by the Core-2.1/Opera 9.5 platform. Opera has one of the most advanced support for Web standards in the industry, and we don't have to care if they are supported yet in another browser. I'm specifically talking about Web standards here, not vendor only solutions, that will cause lock-in. We can use Opera Dragonfly as a showcase of what is possible with the likes of accessibility on the Web. My next mission is to try to get the team to use SVG images in the interface. With SVG we can make the backgrounds of buttons as a reusable element (styleable through CSS SVG profile), use a CSS sprite to reduce the HTTP traffic, and zip as SVGZ to reduce the file size even further than what SVG would already give you. SVG will also allow the button styles to be programatically changed. An example could be detecting the platform and changing the button shapes or colours accordingly. This is not something we've planned yet, but is certainly something that is possible when we don't have to care about IE.

Opera Dragonfly is just over a week old, and I think we can say the early signs are we had a successful launch. It was ready and launched on time, except some gremlins in the server which took a few minutes to sync the development version of the web site with the live site. We got a lot of coverage both in the media and on blogs. We made TechCrunch (and Washington Post), Wired, Digital Web and Ajaxian among others.

I've been trawling the Web for feedback, as part of our team meeting today, to discuss the Beta 1 roadmap. In general the feedback has been positive and understanding that the release was a first alpha. Much of the feedback for feature requests were already on our current roadmap, so I'm hopeful that by the time we hit final, we will have a very useful tool for Web developers. Speed has been a concern, but Opera Dragonfly isn't fully optimised yet, and improvements in Scope, which may come for Beta 1, should help. Currently JSON is not fully supported for everything in Scope, and we find that JSON is much faster currently than XML for communicating between Opera Dragonfly and Scope.

While responding to feedback in the Opera Dragonfly forums, I noticed my first (external) patch. Azamadt Smaguloff (from Kazakhstan) posted a patch on the forums to enable autocomplete in the Command Line. I've not had time to check how this works yet, but this feature is something I've noticed a few people have requested. I think it is fantastic the developers are already interested enough to add to Opera Dragonfly themselves. We don't current take in external check-ins into the official branch, but that doesn't mean we never will.

Things are currently on track for our second alpha, which should (all going well) include inline editing, infrastructure for localisation and bug fixes. Once editing works, that should make the CSS Inspector much more useful for designers.