On Apr 27, 2010, at 11:36 AM, Jeff Morriss wrote:

> Given that the MacOS buildbot (at least the Intel one) is so fast, would 
> it make sense to also build a 64-bit version of Wireshark?

As long as we make sure it works on Leopard before we offer it to Leopard 
users.  The BPF in Leopard, as I remember, had bugs in 64-bit mode and, as

        1) Leopard's compiler built 32-bit by default

and

        2) Leopard didn't ship with 32-bit executables in most cases - most 
significantly, for tcpdump

and

        3) 99 44/100% of Mac OS X applications don't capture network traffic

most of the bugs weren't found until Snow Leopard shipped a 32-bit/64-bit 
kernel and 32-bit/64-bit tcpdump (and, as tcpdump, unlike Wireshark, doesn't 
use a sub-second timeout when capturing, one of them wasn't even found then - 
it showed up with Wireshark when built on Snow Leopard, and was fixed in 10.6.2 
with a libpcap workaround, and un-fixed in 10.6.3 when the workaround was no 
longer being applied due to a change in the way libpcap was built).

We could, I guess, build dumpcap 32-bit only, and just build Wireshark and 
TShark 64-bit.

Also, that would require arranging, somehow, to have both 32-bit and 64-bit 
versions of GLib installed, with the different versions of Wireshark built 
against the different versions.  You could make the *libraries* 2-way fat, but 
making the *header files* work both in 32-bit and 64-bit mode might take a 
little more work.  (Or have separate buildbots.)  For the OS versions on which 
we can make both the 32-bit and 64-bit versions work, we'd probably want to 
lipo them together - along with the supporting libraries - so that there's one 
version that would work on both 32-bit and 64-bit processors.

___________________________________________________________________________
Sent via:    Wireshark-dev mailing list <[email protected]>
Archives:    http://www.wireshark.org/lists/wireshark-dev
Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev
             mailto:[email protected]?subject=unsubscribe

Reply via email to