[Twisted-Python] PB checkers bug?

Steve Freitas sflist at ihonk.com
Thu Jul 28 01:23:00 EDT 2005


Okay, good to know I'm not crazy. I'll file a bug with a patch to change
the docstring.

I'll just put up with a hashed login, as I'm getting impatient to move
on with my project; it'll still allow me to throw a bunch of metadata
into the username string, which gives me everything I need.

Thanks,

Steve

On Thu, 2005-07-28 at 01:09 -0400, Jp Calderone wrote:
> On Wed, 27 Jul 2005 21:22:50 -0700, Steve Freitas <sflist at ihonk.com> wrote:
> >Hi all,
> >
> >I didn't get a reply to my earlier post about inventing mad crazy
> >checkers, so I decided to just use the IUsernamePassword interface, but
> >I'm stuck, because it appears PB won't allow me to use only that
> >interface. If I try, it dies like this:
> >
> 
> PBClientFactory's login method is a rotten lie.  It /does not/ take an IUsernamePassword credentials objects, nor does it log in with the server using an IUsernamePassword credentials object.  It takes an object with a username attribute and a password attribute, and logs in with an IUsernameHashedPassword.  Your credentials checker /must/ support checking the latter interface.
> 
> If you want to support some other authentication backend which is incapable of checking IUsernameHashedPassword credentials, you may need to implement an alternate login scheme for PB.  This is not difficult, but consider carefully whether you actually need it or not (because writing crypto code is always risky).
> 
> Jp
> 
> _______________________________________________
> Twisted-Python mailing list
> Twisted-Python at twistedmatrix.com
> http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python
-- 
Steve Freitas <sflist at ihonk.com>





More information about the Twisted-Python mailing list