(In reply to comment #25)
> (In reply to comment #23)
> > However, I then reproduced the problem using Mozilla's Firefox 3.5.5 package
> > for Linux, which presumably uses its own bundled libsqlite3.so.
> Ehsan - this looks like it may be a private browsing bug assuming my test also
> passes on Linux, or maybe it's a cookie bug.
This is the code responsible for deleting the cookies:
(In reply to comment #25)
> (In reply to comment #23)
> > However, I then reproduced the problem using Mozilla's Firefox 3.5.5 package
> > for Linux, which presumably uses its own bundled libsqlite3.so.
> Ehsan - this looks like it may be a private browsing bug assuming my test also
> passes on Linux, or maybe it's a cookie bug.
This is the code responsible for deleting the cookies:
<http:// mxr.mozilla. org/mozilla- central/ source/ browser/ base/content/ sanitize. js#134>
For the case of deleting everything, it simply calls nsICookieManage r::RemoveAll, which boils down to: <http:// mxr.mozilla. org/mozilla- central/ source/ netwerk/ cookie/ src/nsCookieSer vice.cpp# 862>.
This code looks pretty sane to me.
The thing I'm not sure about is whether your test actually represents the cookie service's db queries precisely. Dan can probably comment on that.