Huge thanks to our Platinum Members Endace and LiveAction,
and our Silver Member Veeam, for supporting the Wireshark Foundation and project.

Wireshark-bugs: [Wireshark-bugs] [Bug 8313] new dissector for SML protocol

Date: Thu, 14 Feb 2013 04:00:12 +0000

Comment # 4 on bug 8313 from
A few more minor questions and issues:

- Why did you write add_length_data()? Instead of "add_length_data(length,
data)" can't you just do "*length + *data"?

- The last argument to any call to proto_tree_add_item should be one of the
ENC_ values (ENC_NA, ENC_BIG_ENDIAN, ENC_LITTLE_ENDIAN). I'm not sure which
version of Wireshark you've been developing against, but we haven't used
TRUE/FALSE for those fields in ages. (Similar point applies to
proto_tree_add_*_format calls).

- Modelines are great, but you don't need them at the top *and* the bottom of
your file. Just the bottom ones are fine. The link I provided is unclear on
that point, I admit; sorry about that.

- You mention in one of your comments that it's "like BER". I know Wireshark
has some BER support functions already (though I myself am not at all familiar
with them). Would using them help your dissector? See
epan/dissectors/packet-ber.h

- You seem to be using integers for booleans in a few places (eg 'close1' and
'close2', line 2361). Wireshark uses glib, which provides a 'gboolean' type
that can be assigned TRUE or FALSE. This may simplify some parts of the code.

- You have some magic numbers in switch statements (lines 2492, 2497, etc). It
would be better to #define them as usefully named constants, then use those
defines in the switch statement and in the corresponding value-string (line
153, etc.)


You are receiving this mail because:
  • You are watching all bug changes.