Howdy,
About a year ago r41216 fixed bug 3315, so that fragments which contributed to a reassembled PDU that matched a display-filter would be exported along with the filter-matching PDU's final frame. I.e., suppose I have a display-filter of 'sip' and frame #3 is the only frame displayed because it matches the filter; but it was really the reassembled PDU of frames #2 and #3 because they were fragmented IP or separate TCP segments... then both frames #2 and #3 would be exported even if I choose to only export "displayed" frames.
That fix was great, but only done in Wireshark not tshark. Even in two-pass mode (opt '-2'), tshark won't print/write the fragments which contributed to the reassembled PDU. Thus there're bugs 8223 and 8101, and http://ask.wireshark.org/questions/18975
I have a patch to fix it, using the same function code as Wireshark. It only works in two-pass mode, since one has to do two passes to accomplish it; but when enabled it changes how the original '-2' two-pass mode displays its output.
So the tricky thing, and the reason for this email, is whether it needs a new option such as '-Y', or even '-Y <display filter>'. I currently have it as a new '-Y <display-filter>', which automatically does a two-pass analysis and ignores the -2 option, and prevents the user from doing both -R and -Y at the same time by error-ing out. I should probably have it error-out if the user indicates '-2' at the same time, too. It leaves the current behavior of '-R' and '-2' unchanged when they're used as before.
Does anyone have any preference/better-idea for how to indicate this new option?
-hadriel
