Re: ftconfig.h

[David Turner]

Hi Antoine,

  I'm forwarding this to the devel list..

Antoine Leca a écrit :
> 
> David Turner wrote:
> >
> > > > Given that <limits.h> is a standard file, I don't think we should remove
> > > > it from this public file. However, I'd appreciate any comments on this
> > > > specific issue..
> 
> Furthermore, it is quite difficult to replace <limits.h> with anything else.
> 
> Perhaps you can point out (in the FAQ) that if the compiler complains
> about lack of <limits.h>, there is a publicly available program named
> (IIRC) enquire.c that is able to produce a very correct substitute for it.
> 
> I believe it is included in the sources for gcc, but should be available
> independantly. Anyone has an URL?
> 
> > > I would say leave it as standard files tend to be protected with
> > > #ifndef _LIMITS_H_
> > > #define _LIMITS_H_
> > > and the like, but I am no expert!
> 

Well, we can also propose to simply replace "ftconfig.h" with a custom
version that defines FT_SIZEOF_INT a FT_SIZEOF_LONG correctly :-) That's
probably a lot simpler than using "enquire.c" or Autoconf/Automake to
automatically detect the integer sizes..

> These are not adequate where you #include the file, but rather inside the
> include file itself, to prevent multi inclusion.
> Also, the very label of the #define (here _LIMITS_H_) is very far from
> being standardised (take a look at GCC headers to have an idea...)
> 
> > Well, the real problem is to know wether <limits.h> is available with
> > non-ANSI compilers (you must have an ANSI compiler to compile the library,
> > but not necessarily to use it in your app), or even in, say, C++ ??
> 
> No problem with C++, AFAIK.
> 
Good to know..

> But <limits.h> certainly is an invention of the ANSI C committee, so probably
> it may be absent from pre-ANSI or non-ANSI compilers...
> Which is the reason why I suggest using enquire.c
> 
> Antoine

Thanks for the input,

- David