Comment # 4
on bug 8610
from Evan Huus
The general problem, which should apply to any protocol that has a tap, is that
the tap must happen after full dissection, in order to have access to the
necessary data, but the tap must happen *even if the dissection throws an
exception* in order to count packets correctly.
Perhaps taps should be queued call-back style so that they can be added at the
beginning of dissection (before any potential exceptions) but then are
guaranteed to get run even if dissection throws an exception?
You are receiving this mail because:
- You are watching all bug changes.