TD>
  >                                                       Additionally,
  > I don't know of many beginner programmers who use multiple compilers
  > (especially those who don't have lots of money). So portability
  > across compilers would not seem to be an important issue in this
  > case.
TD>

  It is if they expect to use textbooks or to ask questions here in
  OS2PROG.

  The reason is that VIO is compiler-independent, and works the same for
  *everyone*, both textbook writers and OS2PROG participants alike.

  Whereas compiler-specific libraries such as the functions commonly found
  in  have traps and peculiarities that will catch the unwary.

  If someone asks about putch() on Borland C++, for example, they are
  going to be misled if someone answers based upon experience with putch()
  on Watcom C++ or MetaWare High C++.  Confusion results.

  But if someone asks about KbdCharIn, then the answer is the same, no
  matter what compiler, or language, is being used.

  The same applies to textbooks.  Many is the "how to program in C++"
  textbook that is written that assumes that such compiler-dependent
  functions as putch() work as they do on the author's C++ compiler, and
  many are the complaints that are made when the example code either
  doesn't compile or doesn't work when compiled with another C++ compiler.

  ( Many textbooks are particularly shortsighted and narrowminded in this
    area, in fact.  It always surprises me that people who use textbooks
    and find that the examples don't work with their C++ compiler, their
    make program, or their linker, then complain about the tools.  The
    fault is clearly with the textbook, not the tools.  )

TD>
  > As for the suggestion that using system calls is the most portable
  > solution, I have to point out that the method of calling 16 bit (and
  > all VIO calls are 16 bit) code from otherwise 32 bit code, does in
  > fact vary according to compiler.
TD>

  Nevertheless, the source will always be the same across all platforms,
  and the behaviour of the system call will always be the same.  What more
  do you want ?

  ( The only known thunking problem, Paul Edwards' Borland structure
    thunking bug, doesn't apply to VIO, since none of the VIO calls return
    a 16:16 pointer to a structure.  )

TD>
  > While many of the compilers' library calls might be one to one
  > translations of VIO calls, some might be a little less straight
  > forward, [...]
TD>

  Whilst the above is quoted from your discussion of reading the library
  source as a guide to using system calls, it does, ironically, prove the
  point that I am making.  Not all libraries implement areas such as
   in the same way, and the programmer who uses  rather
  than using the system API is asking to be both confused and frustrated
  at the end of the day.

  > JdeBP <
___
 X MegaMail 2.10 #0:
--- Maximus/2 3.01