Bug 21957
| Summary: | WebKit doesn't know how to recover from kCFErrorDomainCFNetwork: 302 very well | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Matt Bishop <matt> |
| Component: | WebCore Misc. | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED INVALID | ||
| Severity: | Normal | ||
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | All | ||
| OS: | OS X 10.5 | ||
Matt Bishop
If I visit a site that breaks the connection such that it generates a error "kCFErrorDomainCFNetwork: 302", it takes a while, sometimes never, for that site to start rendering again. A while could be days. The site itself could be working fine, as I can open the same site in Firefox and curl just fine.
The only workaround I have found is to:
1. Empty Cache
2. Delete the site.webhistory file from ~/Library/Caches/Metadata/Safari/History
3. Remove the site's Cookies in Preferences
4. Restart Safari
It may be that the last step (removing the cookies) is the deciding factor of the workaround.
I have seen this in Safari 3.1.2, GWT Hosted Mode (uses WebKit + SWT) and the latest nightly build as of this bug filing.
Searching the web finds many, many people encountering persistent kCFErrorDomainCFNetwork errors like this, so it appears to be a widespread problem.
I can generate it pretty easily if I debug GWT in Hosted mode, hit a breakpoint, then open the page I am debugging in a safari browser instance.
Interestingly, if I stand up Tomcat and replicate the path that GWT uses (http://localhost:8888/my.module.name/index.html) with a plain HTML file, Safari still shows the 302 error.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Mark Rowe (bdash)
For what it's worth, error 302 in kCFErrorDomainCFNetwork is kCFErrorHTTPConnectionLost.
Matt Bishop
I have narrowed the workaround down to just deleting the cookies for the website (in my case localhost/...) and restarting all instances of WebKit. I know, doesn't make sense, but it's what I am finding.
I'm not sure what it is about the cookies I am saving. I will keep trying to narrow this down.
Matt Bishop
The bug is in Tomcat, not WebKit. This WebKit issue can be closed.
FYI, I filed a bug with Tomcat: https://issues.apache.org/bugzilla/show_bug.cgi?id=46125