The first successful smartphones

9. The first successful smartphones

This chapter covers highlights of the first successful projects to create Symbian-powered smartphones. Some of the lessons learned during these projects cast long shadows over later Symbian projects – both good shadows and bad shadows.

Extending EPOC

What’s the difference between a PDA and a smartphone? Most of the people involved in the creation of Symbian would have said, at the time, “Not very much”. If you have an efficient, fit-for-purpose operating system for a PDA – such as EPOC – then it shouldn’t take that much effort to adapt it for smartphones. That was the theory. That theory was especially believed to be true if the operating system already had rich support for multiple communications channels.

Hindsight shows that this widely held view significantly underestimated the work required. It turned out to be much more difficult than expected to extend EPOC to fully support the smartphones and communicators that the Symbian licensees wanted to create. Here are some examples:

  • Some of the devices had hardware that lacked an MMU (Memory Management Unit) – whereas EPOC was written on the assumption that an MMU would always be present, to prevent badly written applications from over-writing each other’s in-memory data
  • EPOC’s support for communications was immature – strong in some places, but relatively untested in others
  • There were holes in the real-time capabilities for EPOC – capabilities that were meant to enable the operating system to attend to incoming high-priority interrupts regardless of what else it was doing at the time; in reality, EPOC could not always give the guarantees required for software such as fax receive and the GSM wireless signalling stack
  • Psion PDAs were “always on”, with the screen blanking and the processor being suspended when the device was inactive, rather than the software being fully shut down; in contrast, phones were more frequently powered right off, and therefore, were more frequently switched on again; this made it much more important that the “cold boot” start-up sequence executed quickly (in less than a minute); however, the design of EPOC did not optimise that start-up sequence
  • Psion PDAs pre-supposed on-board RAM (Random Access Memory) which continuously received power from one or other on-board battery; some smartphones lacked this mechanism and instead required that data be written to an alternative, “Flash” form of memory, which had very different operating characteristics.

In addition to these low-level extensions and adaptions of EPOC, Symbian engineers needed to give attention to:

  • Supporting a fuller range of colours (whereas screens of the initial Psion PDAs just displayed shades of black, white, and grey)
  • Supporting alphabets with larger numbers of characters – such as those used in China, Japan, and Korea
  • Supporting text that flowed from right-to-left, such as Arabic and Hebrew
  • Creating several different user interface frameworks (tasks that turned out to be particularly challenging).

Each separate task could be carefully planned, as an individual piece of work, but it was the interaction of different changes that proved harder to predict and manage – especially when each different licensee product team had their own favourite change requests that they insisted should be treated as “high priority”. The greater the number of changes, the greater was the likelihood of surprise interactions.

Stepping back from the detail, the general principle being illustrated here is this: Moving a complex operating system from one field to another can take a lot more time and effort than initially expected, especially when the new field (e.g. smartphones) involves several whole new areas of requirement and functionality. This change becomes even harder to carry out when there are “too many cooks in the kitchen”.

This principle goes some way to explain the delays in the launches of the first generation of Symbian-powered smartphones – delays that were very frustrating to people at the time, but are more understandable in retrospect. But the story of these first few successful smartphones isn’t just a story of delays and disappointment. Far from it: the story also contains a great deal of innovation and creativity – plus enormous amounts of sheer hard work.

Via innovation, creativity, and (yes) perspiration, the technical problems listed above were all solved in due course, resulting in a software system that was fit-for-purpose, not just for PDAs, but also for a wide variety of designs of communicator and smartphone. By that time, the name “EPOC” was being deprecated, in favour of the term “Symbian OS”.

Roxette: conception

In later years, one design would come to increasingly dominate the smartphone industry: the “mono-block black slab”, with few (if any) moving mechanical parts, and few dedicated hardware keys. But in earlier years, there was no agreement on what the winning form factor(s) would be. For example, many companies experimented in different ways with devices that incorporated mechanical flaps, sliders, or hinges. These devices often had two different modes of operation:

  • With the flap or hinge “closed”, the device looked like a standard mobile phone, with a relatively small screen
  • With the flap or hinge “opened”, there was a larger screen and/or an additional keyboard.

That was the basic model followed by the first ever communicator, the GeOS-powered Nokia 9000, launched in 1996. Ericsson adopted a broadly similar idea the following year for their own GeOS powered GS 88 – though that device was never brought to market. Ericsson brought a revised version of the GS 88 plans to Psion Software in early 1998, for a product codenamed “Roxette”:

  • This was designed to be smaller and lighter – comparable in weight and size to many of the existing standard mobile phones
  • There would be no comparisons with the so-called “brick” of the Nokia 9000
  • The device had a movable flip, which doubled as a keypad
  • With the flip closed, the device displayed a small screen (equivalent in height to around six lines of text) and behaved in similar ways to standard Ericsson mobile phones; in that mode, users operated the phone by pressing on the keypad and by a roller button on the side of the phone
  • With the flip opened, the screen more than doubled in size, switched to landscape mode (instead of portrait), and became touch sensitive.

Psion Software had many reasons to like this project – reasons which remained valid as Psion Software converted into Symbian in the middle of that year:

  • The Roxette project was ambitious, but not too ambitious; for example, the screen would have no colour, but would just display shades of grey; also, there would be no GPRS, nor any Bluetooth
  • Timescales were credible: the product was intended for public launch at the CeBIT tradeshow in March 1999, with product shipments by October that same year
  • The processor would run four times as fast as the Series 5, so we expected the software would have good performance.

The project would also have an interesting financial model: Ericsson were ready to pay Symbian significant amounts of NRE (Non-Recurring Engineering) fees, for work done by a dedicated “LPD” (Licensee Product Development) team in support of Roxette.

[ SNIP ]

Recent Posts