Summary: | Reproducible Poof Crash when using search field at mathworks.com | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Benjamin Olswang <bolswang> | ||||||
Component: | New Bugs | Assignee: | Anders Carlsson <andersca> | ||||||
Status: | RESOLVED FIXED | ||||||||
Severity: | Normal | CC: | adele, beau, mrowe | ||||||
Priority: | P2 | ||||||||
Version: | 412 | ||||||||
Hardware: | Mac | ||||||||
OS: | OS X 10.4 | ||||||||
URL: | http://www.mathworks.com/support/ | ||||||||
Bug Depends on: | |||||||||
Bug Blocks: | 9610 | ||||||||
Attachments: |
|
Description
Benjamin Olswang
2005-07-29 02:24:09 PDT
Confirmed with ToT WebKit. Created attachment 3179 [details]
A minimal page that exhibts the crash.
Created attachment 4939 [details]
Don't call onblur when the focus node is being removed.
This patch makes Safari follow Mozilla and not emit onblur when a node that has
focus is being removed from the document.
Would it work to set m_attached to false at the beginning of detach, and then test for that, isntead of
adding an extra state variable?
Also, don't you want <b>&& !oldFocusNode->m_inDetach</b> rather than <b>&& oldFocusNode-
>m_inDetach</b>?
Comment on attachment 4939 [details]
Don't call onblur when the focus node is being removed.
OK, looks like a good way to do this. Having m_inDetach could make some other
optimizations possible as well.
r=me
(In reply to comment #4) > Would it work to set m_attached to false at the beginning of detach, and then test for that, isntead of > adding an extra state variable? > I didn't want to do that because it felt like it could break in a lot of subtle ways. > Also, don't you want <b>&& !oldFocusNode->m_inDetach</b> rather than <b>&& oldFocusNode- > >m_inDetach</b>? Yeah. I removed it when I wanted to verify that another crash was fixed, and when I added it I forgot the !. Thanks for catching it! Comment on attachment 4939 [details]
Don't call onblur when the focus node is being removed.
Anders and I talked about this on IRC. The test *is* backwards. He'll submit a
new patch.
*** Bug 3683 has been marked as a duplicate of this bug. *** |