Horray for throwing large sums of money at threads... in increments of $0.02!
In my eyes, the main problem is speed. The other main issues are variable and program space, but those don't concern me as much. 38Hz is just too slow to do many of the things most of the people I know would like to do, contrary to what Sachiel7 might say (though he (assuming male, sorry) may only be referring to his team, in which case, who knows, he's probably right).
No, it doesn't have to run an embeded Linux kernel. It doesn't have to control a character LCD display. It doesn't even have to have a different enclosure from the Issac32.
In my opinion, Joe Ross put it best, in
this post, when he said that "
Ultimately, what I would like to see is an interface to completely bypass IFI's microprocessor. That way, they can keep using the Stamp 2SX (or upgrade to a P) and many teams can continue to use it. Then, they provide a way for us to plug in our own microprocessor, of our choosing." I believe that we might be seeing an even more gradual shift to this kind of environment for many reasons. I haven't talked to any IFI or FIRST people about the possibilities, but the fact that, for the past two seasons, they have allowed us to construct our own custom circuit boards, when we're only limited by price (which did, I might add,
double, between 2002 and 2003) and availability (DigiKey and FutureFAI as allowed distributors). Perhaps the next milestone may be to allow these seperate circuits to supply PWM signals to the Victors and Spikes (Although, I have to admit that I don't know what kinds of implications this has, offhand. I believe that I've heard that our RCs produce some kind of non-standard PWM signal, *shrugs*, probably totally irrelevant. Just another reminder to hang a scope on one of the RC's PWM outputs...).
Lastly, I've seen so much about using Java/C/C++/Whatever-Else as a main programming language. Microchip's PIC series of microcontroller, Zilog's Z8 series, and probably many others which I've never heard of, already have C based compilers, and give you access to the machine language, which is a heck of a lot more than Parallax does for us, right now (Their tokenizer, to convert what we know of as PBASIC to machine code, is closed source, and it's only a recent advent that they offer it, free of charge, to anyone, though we still have no access to assembly level programming for Basic Stamps).
Blah. I probably have more opinions about this kind of thing, IM or PM me if you take issue with anything I've said, or have any questions.