[Users] [Bug 3169] threaded message list performance issue

[noreply at thewildbeast.co.uk]

http://www.thewildbeast.co.uk/claws-mail/bugzilla/show_bug.cgi?id=3169

--- Comment #2 from Igor Mammedov <imammedo at redhat.com> ---
Created attachment 1371
  -->
http://www.thewildbeast.co.uk/claws-mail/bugzilla/attachment.cgi?id=1371&action=edit
callgrind sample file

Attached a sample taken with callgrind tool at the moment of issue.
I've used kcachegrind GUI tool to explore it.
Let me know if file works for you and you can use it.

There is no nice stack trace. it's called from gdk_event_dispatch()
here is top 2 callers of tree_draw_node():
 73% 356000times gtk_cmtree_node_set_text()
 18%  66000times gtk_cmtree_node_set_foreground()

the most expensive LOC in tree_draw_node():
 90%inclusive  "while (work && work != node)"

which calculates num index inside tree iterating over tree
which leads to computational cost explosion.

-- 
You are receiving this mail because:
You are the assignee for the bug.