Wireshark-dev: Re: [Wireshark-dev] dissector_add_uint() usage
From: Raghavan R <raghavanr68@xxxxxxxxx>
Date: Sun, 27 Mar 2011 20:46:57 -0700
Hi Guy,

On Sun, Mar 27, 2011 at 8:02 PM, Guy Harris <guy@xxxxxxxxxxxx> wrote:
>
> On Mar 27, 2011, at 7:42 PM, Raghavan R wrote:
>
> > I am writing a dissector for a Fibre Channel related protocol and want
> > my dissector to get called based on value of fcct.gstype.
>
> Then you will have to modify the Fibre Channel dissector so that it has a dissector table that uses the value of fcct.gstype to choose which dissector to call.
>
> At this point, I think one of my biggest mistakes was to give dissector tables names that were the same as the name of the field they used, because that causes a lot of people to think you can use *any* named field as a dissector table.  You *can't* do that; registered fields and registered dissector tables are separate, and almost all fields to *not* have a dissector table associated with them (because almost all fields are *not* used in protocol implementations to select which protocol module to hand the packet to).

Aha - this is what I was missing :-) Thanks a ton for clarifiying it.
I think most people would start from the developer guide like I did,
so maybe you can simply put this note in Section 9.2.

Will try this out over the week. I'm hoping I can make my first
contribution to Wireshark soon :)

Thanks,
Raghavan