Wireshark · Wireshark-dev: Re: [Wireshark-dev] [Wireshark-commits] rev 39143: /trunk/epan/dissectors/ /trunk/epan/dissectors/:

[Guy Harris <guy@xxxxxxxxxxxx>]

On Sep 27, 2011, at 12:11 PM, Maynard, Chris wrote:

> I suppose, although I still question the real need for ENC_NA at all.

At this point, I'm willing to see it go away, too.

> Now that being said, changing the "const gboolean little_endian" argument to an "int endian" or better yet, "int encoding" argument

$ egrep encoding epan/proto.h

	...

    const gint start, gint length, const guint encoding);

> and then performing verification could open up other possibilities.  To quote from README.developer:
> 
>                                                   In the future, other
>    elements of the encoding, such as the character encoding for
>    character strings, might be supported.

I need to remove the future tense from that statement.

> If other elements of the encoding are to be added, then we will also likely need to change the definition of ENC_BIG_ENDIAN (and possibly ENC_LITTLE_ENDIAN too) so that they are both single bit assignments,

$ egrep 'ENC_BIG_ENDIAN|ENC_LITTLE_ENDIAN' epan/proto.h
 * So, for now, we define ENC_BIG_ENDIAN and ENC_LITTLE_ENDIAN as
 * the same.  We therefore define ENC_BIG_ENDIAN as 0x00000000 and
 * ENC_LITTLE_ENDIAN as 0x80000000 - we're using the high-order bit
#define ENC_BIG_ENDIAN		0x00000000
#define ENC_LITTLE_ENDIAN	0x80000000

> thus leaving room for other bits to be set, which will give the encoding a different meaning.

$ egrep ENC_ epan/proto.h

	...

#define ENC_TIME_TIMESPEC	0
#define ENC_TIME_NTP		2

	...

#define ENC_CHARENCODING_MASK	0x7FFFFFFE	/* mask out byte-order bits */
#define ENC_UTF_8		0x00000000
#define ENC_ASCII		0x00000000
#define ENC_EBCDIC		0x0EBCD1C0

ENC_ASCII being all zeroes is temporary, and I will be introducing some additional ENC_ values for character codes.