Amen Russ, I couldn't have put it better myself. Whilst we share the same name, I think we also speak the same language.

Erik, take this on board - everything is not so black and white. One could say that to avoid an accident, jump on your brakes. That may work fine in many cases, but what if you have a truck right behind you? You might want to adopt a different strategy in this instance. The same with pre-emptive RTOSes on 8051 architectures - I'm sure there are some instances where it would be a good choice, but overall one would suggest it be avoided due to the various compelling reasons. To say it is 'stupid' is plain false, more correctly, in some circumstances we could say that a 'more optimum solution can be found'. I . like you would avoid using a pre-emptive RTOS on a 8051, but as I've said before, I've seen product that do and the programmer knew exactly what he was doing and it would have been difficult to have it work using co-operative means. The programmer in this instance would copy the stack to/from xram on a 12 clock part. He knew how long it would take and it was real-time enough for what he wanted to do.

I think what Attila is trying to achive is something like Code Architect - you tick the boxes and it spits out the code module. The 'program' has its own file with the various bits of code and depending on what the user has asked for, it picks out what is required and puts it into an output file. Thus, the PC side of things does all the thinking. So, to cope with the differences between 8051 compilers, the PC code asks the user what compiler and can take of of the required changes.