To both Mr. Simms and Mr. S. Williams: ROL have been cleaning up ROS 4 for several years now. They singlehandely turned this macrokernel beast into a fully modular microkernel baby. In fact they converted most (if not all) modular items into 32-bit complaint modules AND removed most bugs out of it (Yes RISC OS 4 is probably the OS with the least bugs available although it may not be most advanced OS anymore). THIS is the PRIME requirement before any serious kernel modifications CAN and must be done. Therefore they have my deepest respect in what they have accomplished. It is indeed a pitty that it took quite long. It also means that RISC OS 4 is technically the BEST basis to build a 32-bit RISC OS from. Besides do we really KNOW whether ROL hasn't got a 32-bit ROS-kernel or is that just an assumption?
As for Castle. It IS a tremendous effort that they in an absolute record time produced a completely new RISC OS machine with many features we all were dreaming of including a 32-bit RISC OS (which they incidentally didn't build themselves). It is a pitty that the machine is financially again out of reach for most users (especially since most important assets were bought out from a dying PACE) . Things are even worsened by the fact that RISC OS 5 is KNOWN to be very unstable compared to e.g. RISC OS 4 adjust.
So both companies have done miracles and I am proud of their respective achievements. It's a shame they now face each other instead of embracing each other.
Also Castle made assumptions that they wouldn't mind providing (read: selling) 26-bit RISC OS 4 to current interested parties. However they also made perfectly clear that IF ppl wanted to have a 32-bit RISC OS they'd have to BUY new machines with RISC OS 5. They certainly WILL not make RISC OS 4 32-bit nor sell a 32-bit RISC OS 4. They would provide certain RISC OS 4 features in a future RISC OS 5 but no garantuees were made regarding future development on RISC OS 4 as far as they are concerned. Ofcourse I can not blame them for this persuasive move to force ppl to buy new machines (provided they'd want a new RISC OS native machine). I think I was probably the only one at the Expo show whom applauded this, indeed, bold move.
To Mr. Lavalin: Sorry but I completely agree to Mr. Kendrick regarding VARPC. In fact I frimly believe that VARPC is much better for software sales than software houses creating windows-native versions of RISC OS applications. Or to say it bluntly: Yes, indeed Sibelius has done MORE harm than e.g. R-Comp selling RISC OS Doom + VARPC or Aprictote Studios selling Prophet+VA5000!
Blame those companies that first force ppl in the ECC to buy RISC PC's and then suddenly to please the yankees not only convert their stuff but also refuses any further involvement into the user base that granted themn their success!