[pnfs] [PATCH] update exchange id reply to draft 9

[Benny Halevy]

FWIW, all other time values are stored in a timespec struct.
I'm not sure where the conversion to native structures should take
place for every use case.  In some other cases we may want to keep
the time value in a nfstime structure in core and when converting
into a timespec, with a 32-bit time_t, we should check for
overflow.

Benny

Talpey, Thomas wrote:
> At 03:36 AM 3/8/2007, Benny Halevy wrote:
>> Marc, this is not strictly related to this patch but why are
>> we defining struct nfstime4 and using it in struct nfs_impl_id4?
>> why not just use struct timespec in core?
> 
> Because a struct timespec will have different size and alignment
> depending on architecture? Not good for marshalling packets.
> 
> Tom.
>