Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!bloom-beacon!adam.pika.mit.edu!scs
From: scs@adam.pika.mit.edu (Steve Summit)
Newsgroups: comp.lang.c
Subject: Re: C/2 under OS/2
Keywords: why are the libraries not provided?
Message-ID: <10131@bloom-beacon.MIT.EDU>
Date: 27 Mar 89 05:49:55 GMT
References: <357@rruxd.UUCP> <8903172140.AA04348@explorer.dgp.toronto.edu>
Sender: daemon@bloom-beacon.MIT.EDU
Reply-To: scs@adam.pika.mit.edu (Steve Summit)
Distribution: na
Lines: 52

In article <357@rruxd.UUCP> wws@rruxd.UUCP (W W Scott) writes:
>I'm trying to compile a C program under IBM's C/2 but the libraries containing
>the standard C functions can not be found....

In article <8903172140.AA04348@explorer.dgp.toronto.edu> flaps@dgp.toronto.edu (Alan J Rosenthal) writes:
>I can't believe it!  C/2 .. half a C compiler!  hee hee hee

This isn't even funny.  There were some sample programs included
with Microsoft's version 5.1 compiler which I thought I'd try
(what a stupid idea).  One #included <os2.h>, and then made calls
to some nonportable-looking timing routines, nowhere to be found,
inexplicably named DosSleep and DosTime.

Whenever people develop new machines, languages, or operating
systems, there is this apparently irrestible urge to invent
completely new ways of doing everything.  ("Let's see, we'll need
routines to open a file; and get the time of day; and write
characters to the screen.  What shall we call them?  How about
OpenUpATextFile, get_time_of_day, and
SCREEN$PUTCHARACTERWITHATTRIBUTEWITHOUTMOVINGCURSOR.  Now we'll
just rewrite all our old programs to use the new routines...")

Here's a tip for all of the simpleminded reimplementors out there
who think they can show off their creativity by thinking up new
names for everything: YOU DON'T HAVE TO DO THIS.  You don't have to
invent whole sets of routines, forgetting, until later, a few
important ones which then have to be crudely wedged in.  You
don't have to invalidate every program written so far, and commit
hordes of programmers to tedious and error-prone neverending
rewriting tasks.  You could (take a deep breath and brace
yourself; I know this is an earthshattering, revolutionaly
concept) use something that already exists.  If you're writing a
C run-time library, and you're contemplating a feature that X3J11
covers, use the ANSI feature.  If it's an OS-dependent or Unixy
thing that X3J11 doesn't cover, but P1003 (Posix) does, use the
Posix feature.  Barring that, use any old Unix feature, or VMS,
or even DOS, for all I care.  But use something that exists, and
might have a shred of experience behind it, and which might be
utilized by a few lines of existing code.  Even if the adopted
solution is compatible with only 1 out of N existing machines, or
operating systems, or languages, or display devices, or whatever,
that's better than the 0 out of N+1 ratio you're guaranteed to
get by adding yet another one.

I'm not saying "stifle creativity."  If you're doing something
brand new, innovate to your heart's content.  But things like
opening files and sleeping for a certain number of seconds are
ancient history; they're old-hat; they're not that interesting.
Go out and make some real progress instead.

                                            Steve Summit
                                            scs@adam.pika.mit.edu