A-ah! It's the coloring rules. I forgot about those!
Thanks Jaap!
Fabrizio
On Jan 24, 2008, at 2:37 PM, Jaap Keuter wrote:
Hi Fabrizio,
The proto_tree pointer will be != NULL when the entire tree is needed.
That is not only the case when the packet is selected and the details
are displayed, but also when filters are in effect. These filter
expressions can only be evaluated when the dissection details are
known.
Even if you don't have a display filter set the coloring rules are
based
on display filters and have the same effect.
So, clear out the display filters and disable packet coloring. Then
you
will get proto_tree == NULL and hopefully faster load times.
Thanx,
Jaap
Fabrizio Bertocci wrote:
All,
I am working on a newer version of the RTPS packet dissector (real-
time publisher/subscriber).
I have noticed that my packet dissector gets called every time with
the proto_tree parameter not NULL, including when the packets are
loaded in memory.
Unfortunately there are some complex packets that require some heavy
computation that should not be necessary unless the packet is
selected. As result of this, loading a file (with those packets)
takes quite a long time...
The documentation said that wireshark calls the packet dissector with
proto_tree != NULL only when the entire tree is needed.
Any idea why? Is it something that has changed over the time? Is it
something I'm doing wrong?
I'm registering the protocol on top of UDP with a simple call to:
heur_dissector_add("udp", dissect_rtps, proto_rtps);
Thanks for any help provided!
Fabrizio
_______________________________________________
Wireshark-dev mailing list
Wireshark-dev@xxxxxxxxxxxxx
http://www.wireshark.org/mailman/listinfo/wireshark-dev
