Windows 10 is coming this summer, on July 29th we’re told. Although Windows CE (currently aka Windows Embedded Compact 2013) will continue to be supported, should we expect embedded projects based on CE – or for that matter on Linux – to rush to the new OS?
The 3 flavours of Windows 10 IoT are Core – which requiring just 256MB RAM looks like the effective CE replacement, and seems to be aimed at the traditional embedded market; Mobile which requires more RAM, and is ARM-only, looks like it will be reserved for traditional Windows Mobile/Handheld (i.e. large OEM, fixed UI) projects, and Enterprise – a replacement for Windows Embedded Industry/Standard – x86 only.
Windows 10 IoT Core is free, or very low-cost (we await clarification on this), requires only 256MB RAM, and will only run on a limited set of targets to begin with, including Raspberry Pi 2, and some Bay Trail Intel boards such as Minnowboard Max. Previews are freely available.
In theory, the great attraction of Windows 10 IoT is that it is “one OS” with the rest of the Windows 10 family, so you can write your managed code application on the PC, and it will just work on the target. It will be interesting to see how this delivers in practice – in particular in terms of performance. Integration with the Azure IoT Hub concept, which offers client agents to streamline connectivity, and supports open connectivity protocols such as Alljoyn are also attractive if your device fits in with the much-hyped IoT vision.
My conclusion is positive, but it will take time to broaden the supported hardware beyond the initial maker-community focus, and doubtless there will be some gotchas that we haven’t yet understood (stay tuned for updates). But Microsoft is committed, and the direction of travel is the right one. Windows CE continues to be popular but owing to fundamental dissimilarity has never caught the imagination of the Windows world at large, which is where it seems Windows 10 IoT Core will ultimately have the potential to deliver its greatest effect. – dp