On Sun, Nov 18, 2012 at 2:09 PM, Martin Mathieson < martin.r.mathie...@googlemail.com> wrote:
> > > On Sun, Nov 18, 2012 at 1:50 PM, Alexander Koeppe <forma...@online.de>wrote: > >> Am 14.11.2012 15:53, schrieb Martin Mathieson: >> > >> > >> > On Wed, Nov 14, 2012 at 3:54 AM, Cristian Constantin >> > <const.cr...@googlemail.com >> > <mailto:const.cr...@googlemail.com>> wrote: >> > >> > On Tue, Nov 13, 2012 at 9:15 PM, Alexander Koeppe >> > <forma...@online.de >> > <mailto:forma...@online.de>> wrote: >> > > Am 12.11.2012 11:41, schrieb Cristian Constantin: >> > >> hi! >> > >> >> > >> I am just a bit confused about the way the tcp graphs are >> > presented to >> > >> the user. >> > >> I am using debian/Version 1.6.5 >> > >> >> > >> 1. tcp. if one selects (for example): >> > >> >> > >> Statistics/TCP Stream Graph/Window Scaling Graph >> > >> >> > >> from the main menu, one gets the graph from the perspective of >> the >> > >> endpoint which is >> > >> represented by source ip address, source ip port of the frame >> > containing >> > >> the tcp segment >> > >> selected in the main wireshark window, right? >> > > >> > > Right. As the source can only tell something about the window >> > available >> > > on it's end in the ACK packages, this selection has being used to >> > graph >> > > the values. It's up to the user choosing the right direction as >> it has >> > > been done with the other graphs (let along the oposite direction >> > of the >> > > values graphed in the other graphs). >> > >> > cristian: yes, but my point was that for sctp this is done in a much >> > more >> > intuitive fashion, which imo is more user friendly. >> > >> > > >> > >> for getting the graph for the other endpoint, I have to close the >> > >> current graph and select a tcp >> > >> segment with the source ip/port of the other endpoint and select >> > again: >> > >> >> > >> Statistics/TCP Stream Graph/Window Scaling Graph >> > >> >> > >> wouldn't have been more intuitive and easier to use (from the >> user >> > >> perspective) changing the >> > >> graph dynamically when another source endpoint is selected in the >> > main >> > >> window? or to present >> > >> such a choice in the tcp graph windows? >> > > >> > > You mean to add a option to the graph control window, fired up in >> the >> > > background to toggle between the direction? >> > > >> > >> > crisitian: yes, something like that. from what I understand from >> > martin this should be available in newer versions. >> > >> > >> > You don't get to flip the direction inside the graph window, but you can >> > now also launch the graph in either direction from the conversation >> > table window (see screenshot). You could find the TCP connection that >> > you are interested in, then click both graph buttons to launch both >> graphs. >> > >> > Regards, >> > Martin >> > >> > >> > >> > >> > >> > >> ___________________________________________________________________________ >> > Sent via: Wireshark-dev mailing list <wireshark-dev@wireshark.org> >> > Archives: http://www.wireshark.org/lists/wireshark-dev >> > Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev >> > mailto:wireshark-dev-requ...@wireshark.org >> ?subject=unsubscribe >> >> Nice, >> >> but when I want to switch between the different graph types, wireshark >> crashes with a segfault. >> >> Do you experience the same or is it more related to a local problem on >> my side? I just had to replace a faulty harddisk with the OS on it. >> >> Cheers Alex >> >> > I just managed to cause a crash by switching to 'throuput' in the > direction where no traffic was transmitted... I'll take a look. > Martin > > > I just submitted a fix for the problem I could see in r46079 (wasn't allocating enough elements for a tput graph, so was going off the end of the array trying to add new ellipses). Martin
___________________________________________________________________________ Sent via: Wireshark-dev mailing list <wireshark-dev@wireshark.org> Archives: http://www.wireshark.org/lists/wireshark-dev Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev mailto:wireshark-dev-requ...@wireshark.org?subject=unsubscribe
