Summary: | SecurityOrigin::copy does not copy m_domainWasSetInDOM | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Adam Barth <abarth> | ||||||
Component: | Platform | Assignee: | Nobody <webkit-unassigned> | ||||||
Status: | RESOLVED FIXED | ||||||||
Severity: | Normal | CC: | beidson, collinj, mjs, sam | ||||||
Priority: | P2 | ||||||||
Version: | 528+ (Nightly build) | ||||||||
Hardware: | PC | ||||||||
OS: | OS X 10.4 | ||||||||
Attachments: |
|
Description
Adam Barth
2008-01-11 18:19:54 PST
Created attachment 18402 [details]
Fixes SecurityOrigin::copy method to copy the security origin
This patch fixes the SecurityOrigin::copy method to copy the security origin.
Comment on attachment 18402 [details]
Fixes SecurityOrigin::copy method to copy the security origin
r=me
Comment on attachment 18402 [details]
Fixes SecurityOrigin::copy method to copy the security origin
After further scrutiny, there's another aspect of this patch that I'm not sure about.
This removes the copy() calls on the protocol and host. Under normal circumstances, copy() is not needed to copy a WebCore::String, but if we intend to use the string on another thread, we do need a copy(). I'm not sure it's safe to remove the copy() here, because I suspect that the caller does use this on another thread.
We either need to prove those copies aren't needed, or create a new version of this patch that retains the calls to copy().
Created attachment 18450 [details]
Makes SecurityOriing::copy make a deep copy of SecurityOrigin
Here's an updated version of the patch that makes a deep copy. I don't know if the SecurityOrigin::copy method is needed, but it's probably a good idea to actually have it make a copy or remove it.
|