I'm going to move forward with this using more of a pure javascript approach 
than AJAX.  To make things simple I'm just going to define a 
CommentAuthenticatorServlet which will generate the html for the comment 
authenticator challenge.  Then i'll just use a little javascript which calls 
the comment authenticator servlet and inserts the html returned.

Now is the last chance to object to this approach and provide alternatives.

-- Allen


On Tue, 2006-01-03 at 13:27, Allen Gilliland wrote:
> Another idea which may be easier in some ways would be to simply setup 
> the html for the comment authenticator with an <img> tag pointing at the 
> url "/roller/CommentAuthenticator".  Depending on what authenticator is 
> configured it could return an image of a math question, or possibly 
> something else.  This wouldn't be quite as flexible as an AJAX solution, 
> but right now we only have the math authenticator, so i'm not sure how 
> flexible we need to be.
> 
> I'm not sure what would be involved in dynamically drawing an image of a 
> math question, but I would hope it's fairly easy.  I'm not sure if this 
> would go against any accessiblity/usability guidlines though?
> 
> -- Allen
> 
> 
> Matt Raible wrote:
> > On 1/3/06, Glenn Nielsen <[EMAIL PROTECTED]> wrote:
> > 
> >>On Tue, Jan 03, 2006 at 12:51:22PM -0700, Matt Raible wrote:
> >>
> >>>On 1/3/06, Allen Gilliland <[EMAIL PROTECTED]> wrote:
> >>>
> >>>>team,
> >>>>
> >>>>just before the holidays, Dave and I discovered a rather unfortunate bug
> >>>>created by the way the current comment authenticators work coupled with
> >>>>how our page caching works.  basically, because the comment
> >>>>authenticators actually embed their html in the page at rendering time
> >>>>the comment authentication cannot work on a cached page because the page
> >>>>will display an old and false cached version of the authenticator html.
> >>>>
> >>>>the problem has technically existed for quite a long time, however it
> >>>>wasn't noticed because in the past we used to have the old /comments/*
> >>>>urls and we also didn't have the option of a non-expiring cache.  now
> >>>>that we allow any page to potentially be cached for a long time we need
> >>>>a way to allow the comment authentication to remain dynamic.
> >>>>
> >>>>my initial thoughts are to use a little AJAX that will call back to a
> >>>>servlet and ask how to authenticate.  i don't have a proposal yet, but i
> >>>>figured i'd ask around first and see if anyone has any other ideas of
> >>>>how to tackle this.
> >>>
> >>>Doing this with Ajax should be easy enough - the question is how
> >>>secure will it be.  If it's JavaScript that can be called, what's to
> >>>stop spammers from calling the same methods?
> >>
> >>I have been playing around with DWR for Ajax which integrates pretty
> >>easily with a J2EE container and supports access controls using Realms.
> >>I expect it would also work with Acegi.
> >>
> >>DWR: http://getahead.ltd.uk/dwr/
> >>DWR Security: http://getahead.ltd.uk/dwr/security
> > 
> > 
> > Yes, DWR does work with Acegi - however, commentors aren't authenticated. 
> > ;-)
> > 
> > Matt
> > 
> > 
> >>Regards,
> >>
> >>Glenn
> >>

Reply via email to