Re: [avr-libc-dev] New include sys/types?

[hutchinsonandy]

Thanks

I'll go with test  patch, but if this runs into trouble or I hit same 
problem again, I make request official  as the least path of resistance.

I think test patch will be fine -  it's used in other tests.

best regards

Andy



----------------------------------------------
Sent from my Dingleberry wired device.


-----Original Message-----
From: Joerg Wunsch <address@hidden>
To: address@hidden
Cc: Andy H <address@hidden>; Weddington, Eric 
<address@hidden>
Sent: Wed, 4 Jun 2008 1:52 pm
Subject: Re: [avr-libc-dev] New include sys/types?



As Weddington, Eric wrote:

> > File can be blank or just define size_t - as that appears to be
> > only relevant bit for AVR.

I just reviewed <sys/types.h> on my FreeBSD machine, and most of the
stuff there is indeed quite Unix specific.  The only other type that
might be of (some) interest for AVRs would be time_t (which is a
signed 32-bit integer, mostly for historical reasons).

> I have no objections to adding such a file.

Neither do I.  The tricky thing is to properly define size_t in a way
that is compliant with the GCC-supplied headers.  I don't really
pretend to follow the entire logic in <stddef.h>.  This header file
will define all these macros that appear to be related to size_t:

#define _BSD_SIZE_T_
#define _BSD_SIZE_T_DEFINED_
#define _GCC_SIZE_T
#define _SIZE_T
#define _SIZE_T_
#define _SIZE_T_DECLARED
#define _SIZE_T_DEFINED
#define _SIZE_T_DEFINED_
#define _SYS_SIZE_T_H
#define __SIZE_T
#define __SIZE_TYPE__ unsigned int
#define __SIZE_T__
#define ___int_size_t_h
#define __size_t
#define __size_t__

The actual type definition used is

typedef __SIZE_TYPE__ size_t;

--
cheers, J"org               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/                        NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)