WRT is Robert Ackland’s baby. A relative newcomer to the Symbian world, Robert joined the Symbian Foundation in May. His responsibility is for technology management, heading up the applications and UI team that looks after the “upper levels” of the Symbian stack as well as its various runtimes: WRT, Java, and native C++. Prior to Symbian, Robert worked for Motorola on Windows Mobile, Android, and its Linux based Java platform, more specifically working in product management for the multimedia and multitasking division. His working career started in management consultancy before being “sucked into mobile and finding it far too interesting to leave.”
When it comes to WRT, Robert sees the Symbian Foundation as cutting through the fog of fragmentation. “The issue with WRT is that everyone has gone ‘wow, this is pretty cool, let’s create our own extensions’. As a result there are many variations developers have to deal with,” says Robert. As a case in point, Robert refers to a recent poll on the Symbian Foundation forum that started with a list of ten web runtime environments, but resulted in a flood of emails that could, according to Robert, have extended the list to over twenty.
The starting point for the Symbian Foundation WRT solution is the current Nokia (S60) WRT. This is incorporated in Symbian^2, and is now fully open source. The next steps will involve the implementation of the Bondi APIs and security model. Robert notes that at least three separate parties have expressed interest in undertaking this work.
“Ultimately, given that we are open source, my goal is to support as many APIs as possible,” says Robert. “If a widget has been created for Palm WebOS, I hope Symbian will understand the packaging and the JavaScript extensions so that the widget can install and run with no porting. The first step in achieving this is support for Bondi and then extending to W3C packaging formats as well as support for HTML 5.” Robert expects these elements will be in the platform by the middle of 2010. Of course devices with these capabilities will not appear immediately.
Another key aspect of the WRT strategy comes with the move to Qt as the native application framework for Symbian. “One of the exciting things about Qt is that it can run widget code within a native application,” says Robert. “This will enable developers to combine the online usability of a widget with the offline capabilities of a native application.”
The adoption of Qt will facilitate a strategy to make the web an integral part of the operating system, opening up a range of opportunities for developers and other stakeholders within the industry. “With the new UI framework it will be possible for our carrier partners to create interactive home screens based on web technology,” says Robert. “They will be able to combine fixed elements, to link to core services, with dynamic elements that can be constantly updated over their network.”
The roadmap also sees the removal, up to a point, of the artificial split between WRT and browser. Today, the APIs available to WRT widgets are unavailable to browsed content. “In the new architecture any part of the operating system or an application will be able to access the rendering and JavaScript engines,” says Robert. “Where security requirements allow, this means an application, a browser page, or a widget can use the same APIs. If there are restrictions they will come from the requirements of the security models in Symbian or Bondi. Practically, this does mean access will vary between types of applications and, for Bondi APIs, the operator, but there will be far fewer limitations compared to now.”
While access to APIs will influence the types of applications built, the adoption of WRT is also very dependant on performance, it needs to be close to that of native or Java applications. As an interpreted language JavaScript will always be at a disadvantage, particularly when the more constrained resources of a mobile phone are taken into account. “We are on the cusp of some interesting developments with JavaScript engines – it’s much the same thing we saw with Java runtime optimisation as Java gained in popularity,” says Robert. “The developments within WebKit and from Google, which is adopting JIT technology to interpret scripts and make them run faster, are promising. We are looking very carefully at these developments to see which will suit our implementation. As most of this work is open source, it’s practical for our contributors to bring the best of these innovations into the platform.” However, Robert points out that these efforts will always be limited by the fact that JavaScript is not compiled: WRT will always be slower than the equivalent native code.
WRT is being marketed as an “application” development environment. While this is somewhat ambiguous, the inference is that WRT can be used the way Java or native is used today. However, as WRT is scripted any code is more or less open. “Any application that needs IP to be invested in the device resident code or has strong security requirements will almost certainly default to a native or Java implementation,” says Robert. “While WRT is becoming more powerful, I believe it will primarily be used for applications that are lightweight, either in the sense that they rely on a server to do the core processing or simply because they don’t embody that much functionality.”
While the relative ease of web development compensates to a great extent for the current fragmentation in APIs and packaging standards for web widgets, is standardisation going to eliminate these issues? “Standardisation tends to lend its self to the lowest common denominator. The goal of Symbian is to provide a platform on which device manufacturers can innovate, so, between these two factors, I believe there will always be some degree of fragmentation. But this is necessary to drive the technology forward,” says Robert. “For us in the foundation the important factor is how quickly we can get adopter companies to donate their innovations back into the platform and through the consensus of our members drive de-facto standardisation as quickly as possible.”
Ultimately the efficiency and effectiveness of WRT as a development option will depend on how the industry chooses to adopt it,” says Robert. “The Symbian Foundation alone won’t fix today’s issues with WRT, but as the industry moves towards a common WRT platform I expect Symbian to be at the forefront.”