[Guy Harris <guy@xxxxxxxxxxxx>]

On Dec 4, 2003, at 8:11 AM, Luis Claudio R. Goncalves wrote:

> Hi!
>
> On Wed, Dec 03, 2003 at 12:39:31PM -0800, Guy Harris wrote:
> |
> | On Dec 3, 2003, at 2:43 AM, Ronnie Sahlberg wrote:
> |
> | >AFAICS your patch reverses the patch I added to packet-smb.c # 1.366
> | >and changes it back into treating this field as a 64 bit integer in
> | >LittleEndian format.
> |
> | The reasons I know of for not just supporting 64-bit integral types 
> in
> | Ethereal are
>
> The suport to 64bit integers present in etherel, using 8-byte arrays,
> seems to be doing the job nicely.

...except for the fact that there's a dispute over whether the code is 
correct or not (and given that Ronnie wrote the 64-bit integer support 
and put in the current code based on captures he's seen, he might be 
right); I suspect that if we had 64-bit integer support the code would 
just to

	offset_high = tvb_get_letohl(tvb, offset);
	offset+low = tvb_get_letohl(tvb, offset+4);
	offset = (high << 32) | low;

and nobody'd be disputing whether that was correct or not.  (The only 
way to dispute that offset_high is at a lower offset than offset_low is 
to assert that the CIFS 1.0 spec is incorrect.)  I'd have to look at 
the 64-bit integer code again to convince myself which approach is 
right; if we used guint64, it would be obvious.