[Guy Harris <guy@xxxxxxxxxx>]

> The reason for separating the support into different options, one for each 
> protocol, instead of just one option controlled by TCP reassembly:
> Well, the only reason I did that was because desegmentation of tcp traffic 
> might consume significant amounts of memory.

If we're going to have separate options for each protocol, I would be
inclined to have the per-protocol options default to being on, and make
sure that if the per-protocol option for a protocol is on, but the
global TCP reassembly option is off, Ethereal and Tethereal behave the
same way that it would if both options were off.

I.e., I suspect people will usually want to flip only one switch to get
reassembly of higher-level PDUs, rather than having to flip a switch for
TCP and a switch for the protocol with the higher-level PDUs.