80-Bus News

  

November–December 1984, Volume 3, Issue 6

Page 31 of 55

Here we have declared a hundred byte array (0 – 99) and set up a character pointer to the start of it. ‘&arrayx[0]’ is the syntax used to return the address of array[0] which is then placed into the pointer ‘point’. For ‘C’ fans everywhere that have noticed that I could have achieved the same by saying ‘point = arrayx;’, fear not, I realise this but it is a good example of ‘C’s addressing capability.

Pointers to functions are also available and are useful in applications that require overlay sections. Executable code is read into an area of memory with a known base address and using function pointers control is passed to it. You are also capable of passing arguments to this function if required.

‘C’ functions

‘main()’ is the name given to the controlling function of any program and always recieves control when the program is entered. Of course you can do what you like as soon as you have entered main, but devotees of structured programming will realise that this is where the controlling calls of the overall program should reside.

In main of the first sample program you can see that 4 functions have been called, namely openit, wrt_nums, printf and exit. You can see where the first two come from, they are in the source listing, but what about the others? Well it may come as a bit of a shock but ‘C’ contains no IO capability of its own, there are no intrinsic functions such as in BASIC or FOR­TRAN. What this means is that if you compiled this program as it stands into a .COM file the linker would declare printf, exit, fopen, fclose, and putw as undefined. Pretty useless on its own eh? But this is where ‘C’ wins over its rivals in some respects. The functions such as printf, which is capable of displaying a formatted string on the screen, have to be loaded from function libraries supplied to you in REL format on your master disk. The required code is extracted from these libraries by using the L80 /s option for example and linked into your program. This means that the runnable program contains only the code necessary and no unused portions.

These function libraries are usually pretty extensive and contain not only the routines defined by the ‘C’ standard but also some which have been added by the compiler writers to make your life easier. You will have guessed from this that you are also able to create your own libraries and place them into your programs in a similar fashion. I have in the course of my duties at Gemini created function libraries for both the SVC and Pluto which are written in assembler.

As a point of interest the ‘C’ standard dictates that arguments passed to a function are pushed onto the stack in reverse order before the function is called. The value of the argument is pushed and not the address of the variable ie. pass by value. So from within your custom written assembler routine you can retrieve these values from the stack and send them out to your SVC or Pluto as data using the Z80 OUT opcode. In practice though, I have noticed that the ECO C compiler only passes integer and character arguments by value and any arguments longer than 2 bytes are passed by address. There are often idiosyncracies involved with the method by which arguments are passed so I’m afraid that there is no substitute for reading the manual.

Page 31 of 55