Hi Ray



On (10 Sep 94) Ray Markham wrote to John Spence...



 JS> I never see a C programme with an include like  #include 



 RM> I have resorted to this a number of times. For example when all the files

 RM> relate to the same set of data, but are functionally different, i.e. input

 RM> and output etc. When the files begin to get too large to be comfortably

 RM> maintained, I will break them down into functionally separate modules and

 RM> #include "input.c" etc. from one main program file. This
gives a place

 RM> where all the header file #include's can take place as well, thereby

 RM> keeping all the code manageable. This is especially important in larger

 RM> projects which can end up being composed of many files.



I think I'd find this approach a bit less manageable. If I came

back to the project after a break, and wanted to #include another

header file in routine X, I'd have to hunt down the "master" file

and include it there.



If the compiler threw up an Undefined Symbol error, I'd be looking

at two files at once to track down the error.



I frequently do a fast syntax check on a C file by telling the IDE

to compile (not link) the current file only. This wouldn't work if

the header file includes were elsewhere.



I managed to think these reasons up in a few minutes, and I'm only

a dabbler in C. Wonder what the real programmers think...



Cheers



--- PPoint 1.86