Wireshark-dev: Re: [Wireshark-dev] [Wireshark-commits] rev 41790: /trunk/ /trunk/: configure.in
From: Guy Harris <guy@xxxxxxxxxxxx>
Date: Tue, 3 Apr 2012 12:56:56 -0400
On Apr 3, 2012, at 12:40 PM, Gerald Combs wrote:

> Bug 6973 contains a patch for a wireless toolbar which uses the Netlink
> library (libnl). This would fix the problem on Linux but it wouldn't use
> common code.

Ideally, the AirPcap and non-AirPcap wireless toolbars should use as much common code as possible (AirPcap might have some options in addition to the channel).

The netlink code should ultimately be stuck behind a common API that could use the *BSD ioctls or the Seekrit Private Framework code on OS X as well as whatever APIs eventually appear in libpcap, but that can be done after it's integrated into Wireshark.

(There should also be support in TShark and dumpcap - I vote for "-Q" as the flag to use to select the channel, as I'd had it in mind for that purpose in tcpdump/dumpcap/TShark/etc.. It's one of the few options that's available in all; one of these days all of the aforementioned programs should start using getopt_long() and bring along their own version of getopt_long() for platforms that lack support - there are both GPLed and BSDLed versions of it, so *shark and tcpdump can both use it.  BTW, I think OS X's "ls" uses almost every letter of the alphabet, both lower-case and upper-case; I wouldn't be surprised if the GNU "ls" did so as well....)

(Oh, and "-Q scan" at some point, to do channel scanning.  But I digress....)