Note, by the way, that if we handle the case of higher-level packets
being made out of pieces of mroe than one link-layer frame, we have to
deal with situations where, currently, it is assumed that a packet
summary line and a protocol tree can be completely generated for a frame
based only on the bytes in the frame, and information gathered from
previous frames, but nothing from *subsequent* frames, such as:
Tethereal, where the summary line or protocol tree is printed as
soon as the frame is dissected;
read filters, where the packet is read in and dissected, and
added to the list of packets in the capture, without reading in
any subsequent packets (note that, in effect, read filters can
be used sort of like capture filters in Tethereal, as well);
an "Update list of packets in real-time" capture;
building the list of packets to be shown in the packet list
pane, where the summary line is generated as the frame is
dissected for the first time.
This is, as you might infer from my mail, a collection of hard problems,
which is why
1) we haven't done it yet
and
2) aren't likely to come up with an implementation for it
quickly.
