Re: [cc65] POSIX i/o

Date view Thread view Subject view

From: Greg King (gngking_at_erols.com)
Date: 2003-01-16 11:56:57


-----Original Message-----
From: Ullrich von Bassewitz
Date: Thursday, January 09, 2003, 04:39 PM
>
> On Thu, Jan 09, 2003 at 02:55:01PM +0100, Piotr Fusik wrote:
> > > Regarding the #defines for seek: The C standard says that these have to go
> > > into stdio.h, so they are in the right place.
> >
> > Shouldn't they be defined both in stdio.h (for fseek) and fcntl.h (for
> > lseek) ?
>
> I see no real need to do that. The standard says that the #defines are
> in stdio.h, so anyone using them just has to include this header file.

Doesn't "lseek.s" need to see them in "fcntl.inc"?

------

We need a prototype for chdir().  (It and rmdir() should be __fastcall__.)

------

> > >>> #define creat(name,mode)  open(name,O_WRONLY|O_CREAT|O_TRUNC,mode)

Is creat() supposed to return an error-value when the file already exists?
Or, is it supposed to replace an old file with a new one?

> Using fcntl.h is the DOS/Windows way, using unistd.h is the Unix way.
> Since I didn't expect fcntl.h to grow, and unistd.h suggests Unix
> compatibility which is not possible, after some discussion,
> fcntl.h was chosen instead of unistd.h.

Are there any prototypes that traditionally go into unistd.h that you wouldn't
put into other headers [sleep() and alarm(), for example]?

----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo_at_musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.


Date view Thread view Subject view

This archive was generated by hypermail 2.1.3 : 2003-01-16 12:00:00 CET