Another good thought provoking article by Mr Naulls !
Much of what he says does make sense, but I am a little concerned about some of the conclusions drawn.
Although having a Linux/BSD based ARM board (or even desktop machine) may gain sales - I am not convinced it will necessarily help RISC OS.
After all the World and their dog probably have Linux (or some variant of Unix) on ARM already - so what is to distinguish a Castle/MD board from the rest ?
As a trojan horse (as CASW1 suggests) perhaps RISC OS might make some headway, but if a firm is producing hundreds or even thousands of development boards - and a RISC OS license fee is due for each and there's no such license required for Linux what odds (at some point) the company involved will simply drop RISC OS (and save a little per board) and simply become a "me too" Linux/ARM board manufacturer.
I think having Linux or BSD on board is no harm - but it of itself while perhaps granting a company some sort of reliable income will not put *any* money into RISC OS or into software development on RISC OS.
Oddly having RISC OS on board *may* add an extra feature bullet point (along with Linux or BSD) that might distinguish such a development board. Trouble is will the manufacturers involved see it as a sufficient enough feature to pay for (and on that may hinge the future of RISC OS itself)