Not really very bad, but it imposes some challenges!


No. The main problem is the lack of any significant stack space. This is why Keil et al (and PL/M before it) have to do all that messing about with overlaying, resulting in inherently non-reentrant functions.

The DPTR has nothing specifically to do with 'C' - it takes the same number of machine instructions to accesses an XDATA byte whether you code in 'C' or assembler.


That is not the fault of the compiler nor even the language - that is the programmer's fault for writing "inefficient" code!
Going through so many levels of indirection will be "inefficient" on any processor.

The problem here is the use of pointers; not structures per se; the following would take no more code than a simple variable assignment:

X.Y.z = ...;

Again, if you tried to write the equivalent in assembler, it would almost certainly end up taking just as much code: think about it - how would you write that statement in assembler? Could you do better that the compiler?


None of this has anything specifically to do with 'C' - it is all about poor use of the tools, and failure to understand the underlying architecture.
It is impossible to write efficient code in any language without a thorough understanding of the underlying architecture.