KEIL should point out, clearly, in their "slices, dices, and makes hundreds of julienne fries" marketing literature, that it doesn't support the Dallas one-clockers. Instead, their own people don't even know it doesn't do that. Yes, it generates "generic" code, but its support for other devices, i.e. the various derivatives many manufacturers sell is pretty limited. I doubt adding a new derivative costs the KEIL folks over the cost of one copy of their software. They don't support the additional hardware in any way. They don't support the peripherals in any way, since they don't simulate them.

I'm not big on 'C' programming, so I don't really miss the compiler. Since I don't care about those derivatives, I don't use the KEIL suite. However, it's time software houses became honest (fat chance!) about what they DON'T do.

There's an industry-wide tendency on the part of software producers to exaggerate the capabilities of their products. There's often a feature or two that really doesn't work, or that only hints at working. Often you have to buy another package in order to get a highly touted feature. Often enough, the one feature that's really important to a purchaser is not available until "the next release." Often their "new, improved" versions don't do anything the previous release wasn't supposed to do, but didn't.

The claim of "support" is pretty meager, in reality. What do they do to give a device "support?" A table of SFR definitions is what you get, right? They give you the ability to use dual data pointers, but do they implement the auto-increment/decrement features? When you use a converter, does it simulate the time it takes to convert, and allow you to provide a value to use as a result? When you use other features, do those features become "just another SFR," or is there some "real" support for them.

How adequate the product, overall, is, is not the subject of this thread. I think the simulator is inadequate, irespective of whether it's the demo version or the "real McCoy." However, I think I have to agree with Jez' initial assessment, that the 2KB-limited version is inadequate for evaluation of the features that ultimately make the difference. Most of those features don't show up in a 2KB task.

RE