@zach @zachary @timb @peekay123 @david_s5
Gents, on the topic of a Spark Core II et al, I really think the topic should get some serious attention. As a prospective commercial user of the platform I started out wildly excited at the time of the Kickstarter project. But first some background.
I have been developing a commercial application (reasonably complex & sophisticated) using the Electric Imp platform, and from a RAM, Flash and wireless stability point of view it has been superb. But, with the Squirrel code being intermediate byte code and thus interpreted, real time processing and interrupt handling became a problem. Also the limitation of not having access to an IP stack (at least UDP) as well as the reliance on the Imp Cloud are all limiting factors. Otherwise, still a superb platform for many tasks.
From the Spark Core side the main attractions have been and remain the open source nature of the project, the detachable umbilical cord to the Spark Cloud and the general excitement and cooperation in the Spark community. However, by the time my Cores arrived reality had sunk in and after very limited experimentation I have shelved my Cores for a couple of reasons.
Memory especially RAM but also Flash. In the era of declining uC (and supporting hardware) cost, it is just not practically feasible to do anything of any real sophistication with a total of 20KB RAM. I believe 64KB must be a minimum maybe 96KB. Flash at least 256 KB but preferably 512Kb.
“Real Time” capability. The current Arduino based execution model does not allow this. In addition the RAM/Flash side stops the use of something like FreeRTOS+???. It is not feasible as a platform with the memory available.
Wireless instability. Also, for any serious use this must be fixed, preferably with FreeRTOS support as well.
The increase in uC spec will add to the cost, I estimate it going up to about $50, but it will increase the opportunity exponentially into the commercial space. Physical size, if it had to go even double to current real estate to enable more IO’s, also still reasonable since the current Core will still be there as a smaller and lower cost alternative.
If you can get the above right, you will create a huge market not just for the Core II/Ultra/whatever but also for breakouts or “baseboards”. I started a process to commission the design of a range of open source mote baseboards (hoping to attract developers to help write supporting drivers for them) to address specific needs I have including:
A) Environment sensing
B) Motion sensing
C) Power measurement and control
D) Human interface
E) General purpose with selected bits of all the above.
My commercial needs remain, and it is burning, but for a significantly higher spec Core. I am happy to have a discussion around specs, cost and an order, should there be serious interest in taking this further.
PS: I apologise for addressing the comment so widely, but hoping to stir up a serious and urgent discussion on this topic.