> Interesting how well ViewFinder works, then, isn't it?
In effect John Kortink has written code to (I assume) intercepts normal OS Video calls and then "re-route" them to the card. Point is that this though well implemented is a bit of a "bodge". It would be better (cleaner) for the OS to handle the hardware abstraction itself as then it wouldn't have to start off steps to use the VIDC and have them "intercepted" by yet more code to re-direct it to a foreign card.
I'd imagine JK would have a lot of work to rejig it to work with a different video card (or if ROL changed RISC OS in anything other than a trivial way).
MD obviously took the view that *if* the hardware looked exactly like a VIDC then no real change to the OS would be required (which is a fair enough approach and if done correctly could have speed benefits). VF coded round some of the RISC OS limitations (but this task would have been easier had the OS allowed it in an easier manner, such as by using a HAL as Iyonix RISC OS 5 does).