Hi,

A heads-up for such a scenario: If your gadget definitions are also on
https servers, and they use user-preferences, there is an issue that
Shindig currently tries to use gmodules.com to render the user prefs.
However, gmodules does not support https :(

cheers,

Niels


 
On 01/17/2011 01:17 PM, Xandeco, Maxwell wrote:
> unfortunately it's not my choose, it's a requirement here, everything must be 
> over a secure server with sso.
>
> Maxwell
> _______________________________________
> De: Christiaan Hees [[email protected]]
> Enviado: segunda-feira, 17 de janeiro de 2011 8:35
> Para: [email protected]
> Assunto: Re: Requesting Gadget Metadata from a Secured URL
>
> Why is your gadget spec xml behind SSO anyway?
> I think in general you should design your gadgets in a way so the spec
> can be public.
>
> On Sat, Jan 15, 2011 at 2:05 AM, Ryan J Baxter <[email protected]> wrote:
>> So the server Shindig is running on does not require the user to log on?
>> Why not for the user to log on when accessing the server running Shindig
>> as well?  That way when to access other URLs that require to log in, there
>> won't be a problem.
>>
>> -Ryan
>>
>> Email: [email protected]
>> Phone: 978-899-3041
>> developerWorks Profile
>>
>>
>>
>> From:   Maxwell <[email protected]>
>> To:     [email protected]
>> Date:   01/14/2011 06:33 PM
>> Subject:        Re: Requesting Gadget Metadata from a Secured URL
>>
>>
>>
>> The domain does not matter here, look I my server has a sso project, that
>> handles all projects in the server, pretty normal stuff here, when i call
>> any url for the first time, my server see that i'm not authenticated, so
>> execute the login action, after that all browser request goes with a
>> cookie
>> (sso cookie), that is also pretty normal, if I try to access my xml spec
>> over a secured url using the browser i can see the file, because the
>> browser
>> is authenticated and send the cookie, everything is fine here.
>>
>> When I ask to shindig render a gadget to me, it does a request, at this
>> moment it's not a user request (browser request). shindig does another
>> request, and do not pass the original user request object
>> (servletRequest),
>> so my server does not allow shindig to access the resource (xml).
>>
>> So I'm trying to figure out a way to do that, but for what i see, i have
>> to
>> rewrite all "call chain" pf the request, from the Servlet to the
>> HttpFetcher
>> to allow pass original request through.
>>
>> Why shindig assume that all server is public?
>>
>> Imagine that igoogle do not let you to see and get the gadgets spec,
>> unless
>> your logged with your account and pass some cookie to it, how you are
>> going
>> to do that?
>>
>>
>> Thanks.
>>
>> On Fri, Jan 14, 2011 at 9:20 PM, Henry Saputra
>> <[email protected]>wrote:
>>
>>> Hmmm are you saying  that Shindig is put in different domain/ url so
>>> the request come from client will go to different URL without cookie?
>>>
>>> - Henry
>>>
>>> On Wed, Jan 5, 2011 at 10:09 AM, Xandeco, Maxwell
>>> <[email protected]> wrote:
>>>> Hi guys,
>>>>
>>>> We have to use a SSO secured server, that means all apps in the
>> container
>>> will be covered by the SSO system, even the gadgets spec XML.
>>>> My server uses a user cookie sent by browser to authorize the access,
>>> it's a simple SSO system, the add/render gadget flow it's basically:
>>>> Browser --> RpcServlet --> JSONRpcHandler --> Processor -->
>>> GadgetSpecFactory --> RequestPipeline --> HttpFetcher
>>>> Shindig does a new request, with any association with the original
>> client
>>> request (that has all cookies necessary to pass over security handlers),
>>> it's just like try access a url without login, so i got a 401 error,
>> it's
>>> pretty easy to replace implementations on shindig using google-guice,
>> but
>>> the only way i see here, is rewritten all classes involved in the
>> process,
>>> because after JSONRpcHandler the original request it's not passed
>> through.
>>>> How do you guys normally handle that, you always put spec in public
>> urls?
>>>> Cheers.
>>>>
>>>
>>>
>>> --
>>> Thanks,
>>> Henry
>>>
>>
>>
> >

Reply via email to