Comment # 8
on bug 11980
from Peter Wu
Theory and practice do not seem to match, either priming or doing the
dissection with additional primed fields is expensive. I am inclined to keep
the previous patch as the architectural implication is nice (ui -> dissection)
and have pushed a patch that should address the performance issue.
I am also tempted to mark dynamically changing coloring rule columns and tree
items as unsupported for now. Otherwise the whole list needs to be re-dissected
after changing a single coloring rule... probably not worth it.
Coloring rules in display filters are also not supported because it needs
additional work in detecting whether it is in use in a dfilter (should be easy)
and additional complexity to optimize (new: first calc while reading packets;
avoid second calc when adding it to the Qt packet list; but keep calculation
when a packet is selected and the full tree needs to be known)
You are receiving this mail because:
- You are watching all bug changes.