Huge thanks to our Platinum Members Endace and LiveAction,
and our Silver Member Veeam, for supporting the Wireshark Foundation and project.

Wireshark-dev: Re: [Wireshark-dev] epan_t and capture_file

From: Michael Mann <mmann78@xxxxxxxxxxxx>
Date: Fri, 14 Apr 2017 19:26:58 -0400
I agree with the suggested name changes, but I still don't think the functionality is necessary.
 
The timeline view (in its current form) seems to look very similar to the new (Qt only) "packet/color summary scroll bar" to the right of the packet list (not sure of its official name).  That needs some level of dissection to determine the color of each packet to draw the scrollbar.  I don't believe it uses taps, but it did not put any additional requirements/code into dissectors.  I would suggest modeling the timeline view after that.  I don't know enough about Qt to make more specific suggestions, but if someone provided the classes used in the Qt GUI code for the scrollbar, I might be able to copy/paste/hack the timeline view to look similar.  The timeline view is certainly a cool feature, I just think it needs some architecture work to be submitted.
 
This is an interesting problem because I'm not sure where the "data crunching" should really be done (GUI vs dissector).  For statistics dialogs, I think more data computation is done in the GUI (taking information from taps).  It doesn't appear to me that the dissector does anything with the computation except to feed it to the GUI, so I don't think it would be bad if it was moved out of the dissector (and remove the need for all of the epan APIs added just for this feature).
 
 
-----Original Message-----
From: Guy Harris <guy@xxxxxxxxxxxx>
To: Developer support list for Wireshark <wireshark-dev@xxxxxxxxxxxxx>
Sent: Fri, Apr 14, 2017 5:13 pm
Subject: Re: [Wireshark-dev] epan_t and capture_file

On Apr 14, 2017, at 2:02 PM, Simon Barber <simon.barber@xxxxxxxxxx> wrote:

> The timeline needs to know whether every packet includes timing information. If not, the timeline should be hidden. This is only determined after all the packets have been through a first dissection pass.

Then there needs to be a mechanism by which the timeline GUI code can inquire whether every packet includes timing information.

The change to allow dissectors to share per-capture data (not "private" data, "private" means you *don't* share it, as per Michael Mann's comment) would be a mechanism to do that.

For now, just change "private" to "per_capture" in the function and variable/field names and to "per-capture" elsewhere, and wait for it to be approved.
___________________________________________________________________________
Sent via: Wireshark-dev mailing list <wireshark-dev@xxxxxxxxxxxxx>
Archives: https://www.wireshark.org/lists/wireshark-dev
Unsubscribe: https://www.wireshark.org/mailman/options/wireshark-dev
mailto:wireshark-dev-request@xxxxxxxxxxxxx?subject=unsubscribe