NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/gnu/gcc/help/3019

Internet Message Format | 1993-01-24 | 4.1 KB

Path: sparky!uunet!stanford.edu!agate!usenet.ins.cwru.edu!magnus.acs.ohio-state.edu!cis.ohio-state.edu!candidus.ma30.bull.com!drg
From: [e-mail] (Daniel R. Guilderson)
Newsgroups: gnu.gcc.help
Subject: [biz.sco.general] Re: gcc 2.3.2 on SCO 3.2.4 (__STDC__)
Date: 24 Jan 1993 13:47:49 -0500
Organization: GNUs Not Usenet
Lines: 70
Sender: [e-mail]
Distribution: gnu
Message-ID: <[e-mail]>

I know nothing about the ANSI C standard, I just know that RMS has
said that he designed gcc so that it would have __STDC__ defined as 1
on all the platforms that it runs on.  I don't have time to find and
read the standard so maybe someone who knows can tell me exactly what
the function of __STDC__ is.  Attached is a message from
biz.sco.general in which a man claims that __STDC__ should be 1 only
when the compiler and the libraries are all ANSI conformant.  He also
claims that __STDC__ should be defined as something else (like 0) when
only the compiler is ANSI conformant.  This seems like a confusing and
useless way to do things.

Please ignore the fact that SCO claims their libraries and header
files *are* ANSI conformant.  I just want to know what the heck
__STDC__ is supposed to mean.

------- Start of forwarded message -------
Newsgroups: biz.sco.general
From: [e-mail] (Steve Bleazard)
Subject: Re: gcc 2.3.2 on SCO 3.2.4 (__STDC__)
Organization: Robobar Ltd., Perivale, Middx., ENGLAND.
Date: Sun, 24 Jan 1993 11:15:14 GMT

In article <[e-mail]> [e-mail] (Daniel R. Guilderson) writes:
>In article <[e-mail]> [e-mail] (Stuart Lynne) writes:
>
>   Saying it is wrong, is well... definitely wrong.
>
>I say it is wrong because gcc was designed to have __STDC__ defined as
>1.  I don't remember exactly which program it was but I once ran into
>a problem were defining __STDC__ to be 0 caused compilation problems.

This is not as cut and dried as it appears.  What does __STDC__ mean?  The
standards appear to say (and I'm using P J Plauger & Jim Brodie's Standard
C book as my reference) __STDC__ is 1 if and only if the ENVIRONMENT is
ANSI.  That is the library and the compiler.  Library functions other than
the ones defined in the standard are definitely out.  Before the standard
was accepted __STDC__ was used to indicate an ansi conformant compiler (and
in some cases not so conformant - thankyou Microsoft!) rather than an ansi
environment.  This has resulted in a lot of code that uses __STDC__ to mean
a conformant compiler, not environment.  This causes a problem for SCO.
On the one hand they want to produce a conformant environment - so I can
check that my program will compile on all conformant systems (including
non-unix) - and on the other hand there is a lot of code in the world that
uses __STDC__ to detect an ansi compiler.  There solution is fairly good:
The standard says __STDC__ can be either another number or not defined if
the environment is not ansi.  Every program I have ever seen that depends
on some ansi feature uses #ifdef.  So defining __STDC__ to be 0 solves
the problem for nearly all cases.  Only in the rare case where __STDC__ is
tested against 1 (or 0) will there be a problem.  The headers can now
say #if !__STDC__ where functions are defined that are not part of the
standard.

>   If you want a GCC compiler that doesn't then fine. Go for it. But
>   don't complain to SCO and ask for support for the munged header files.
>
>On the otherhand, if you define __STDC__ to be 0 don't complain to the
>FSF and ask for support of programs which don't compile correctly.
>
I don't think this is an FSF issue, FSF produce a compiler that is ansi
conformant, they do not yet produce an ansi environment. One can argue
that they should define __STDC__ as 0, leaving it up to the CPP_SPEC to set
it to 1 when an ansi environment is being used.

Steve
--
[e-mail]        | Phone:  [phone] x153
Snr Software Engineer, Robobar Ltd. | Fax:    [phone] (G3)
22 Wadsworth Road, Perivale.        |
Middx., UB6 7JD ENGLAND.            | ...!ibmpcug!robobar!steve
------- End of forwarded message -------