[PATCH] kcookiejar

[Waldo Bastian]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Thu May 20 2004 23:07, Ralf Holzer wrote:
> On Thursday 20 May 2004 04:38 pm, Waldo Bastian wrote:
> > > But loadConfig() only sets the domain-specific
> > > advice for the domain without the prepended dot.
> >
> > That depends on the configuration I would think.
>
> You can try this with slashdot.org. Whatever you enter into the
> site-specific policy dialog (slashdot.org or .slashdot.org), kcookiejarrc
> always stores the domain without the leading dot.
>
> > > This causes findCookie()
> > > to skip over the cookie, because the advice is set to KCookieDunno
> > > (line 348).
> >
> > ?? Why would it skip over the cookie when the advice is KCookieDunno?
> > When the advice is KCookieDunno it will look at the next domain to see if
> > that specifies a policy.
>
> Because the actual domain in the cookie is stored as ".slashdot.org", but
> kcookiejarrc contains "slashdot.org". I added some debug output and saw
> that the ".slashdot.org" entry is set to 0 (KCookieDunno) and the
> "slashdot.org" entry to (1) KCookieAccept. But line 348 only sets the site
> policy, if it is something other than KCookieDunno. The next if statement
> then rejects the cookie for ".slashdot.org" , because the global policy is
> set to Reject (line 354).
>
> > I'm not convinced that it is correct and if it is than there is probably
> > a better way to do it.
> >
> > Can you post the relevant section of your kcookiejarrc file and say on
> > which particular URLs you experience the problem you mention in BR80501?
>
> My kcookiejarrc is attached. It doesn't matter which URL, but you can
> easily try it on slashdot.org. I have the global policy set to reject. Then
> I add a site-specific policy for "slashdot.org" and set it to Accept. The
> cookie never gets stored this way. So I switched the global policy to "Ask"
> in order to store the cookie. Once the cookie is stored, I switch back to
> "Reject".

The problem is that leading dots are stripped.

r1.11 of kidna.cpp is responsible for that. Seems Thiago still hasn't fixed 
that.

Cheers,
Waldo
- -- 
bastian at kde.org  |   Novell BrainShare Europe 2004   |  bastian at suse.com
bastian at kde.org  | 12-18 September, Barcelona, Spain |  bastian at suse.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQFArh0QN4pvrENfboIRAmy6AJ0XFjOdvwu/yi+CoNjQSEo0wxtBgACfUC35
4AGEq6EeVjCdL/1Go6CBlxo=
=9GNE
-----END PGP SIGNATURE-----