Michael Mann
changed
bug 2412
| What |
Removed |
Added |
| Attachment #1632 is obsolete |
|
1
|
| Attachment #1633 is obsolete |
|
1
|
| Attachment #1634 is obsolete |
|
1
|
| Attachment #1635 is obsolete |
|
1
|
| Attachment #1636 is obsolete |
|
1
|
| Attachment #1637 is obsolete |
|
1
|
| Attachment #1638 is obsolete |
|
1
|
| Attachment #10664 Flags |
|
review_for_checkin?
|
Comment # 11
on bug 2412
from Michael Mann
Created attachment 10664 [details]
New expert info architecture
Here is a start to what I believe to be the desired framework. It's a hybrid
implementation (read: copy/paste) of hf items and preferences. This
implementation registers the expert info, but only just uses it to look up
grouping and severity. I figured just getting itemization in place was the
more important thing for now, while keeping backwards compatibility. I think
expert_add_info_format should be able to be deprecated in favor of
expert_add_info and expert_add_info_format_text. I would prefer the
expert_add_info_format_text functionality to be called by the
"expert_add_info_format" function since it's more inline with existing naming
conventions, but I didn't want to rename it in all of the existing dissectors.
One of the main items up for review is the API function names and structures
(for example, perhaps "ei" prefix instead of "exp"?). I want to get this
solidified before scripts are written (volunteers welcome! :) ) to take
advantage of it for all dissectors.
Another review item would be when to call expert_init/expert_cleanup.
Currently it's being called more than once, which doesn't allow the "itemized"
architecture to work because the registered item array is getting cleared out.
I believe this is related to its "hf items", but don't know enough of the
internals to know where to move them to while the transition to "itemized
expert info" is done. Would a simple global "initialized" flag be good enough?
You are receiving this mail because:
- You are the assignee for the bug.
- You are watching all bug changes.