portals-jetspeed-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Santiago Gala <sg...@hisitech.com>
Subject Re: cvs commit: jakarta-jetspeed/src/java/org/apache/jetspeed/portal/security/portlets PortletWrapper.java
Date Mon, 03 Mar 2003 17:24:11 GMT
David Sean Taylor wrote:
> On Monday, March 3, 2003, at 08:10  AM, Santiago Gala wrote:
>> taylor@apache.org wrote:
>>> taylor      2002/09/17 08:44:00
>>>   Modified:    src/java/org/apache/jetspeed/portal/security/portlets
>>>                         PortletWrapper.java
>>>   Log:
>>>   Added method to get around cast firewall.
>>>   Plan to convert this code to properly use Java Proxies.
>>>     Revision  Changes    Path
>>>   1.13      +7 -2       
>>> jakarta-jetspeed/src/java/org/apache/jetspeed/portal/security/ 
>>> portlets/PortletWrapper.java
>>>     Index: PortletWrapper.java
>>>   ===================================================================
>>>   RCS file:  
>>> /home/cvs/jakarta-jetspeed/src/java/org/apache/jetspeed/portal/ 
>>> security/portlets/PortletWrapper.java,v
>>>   retrieving revision 1.12
>>>   retrieving revision 1.13
>>>   diff -u -r1.12 -r1.13
>>>   --- PortletWrapper.java    3 Sep 2002 15:02:13 -0000    1.12
>>>   +++ PortletWrapper.java    17 Sep 2002 15:44:00 -0000    1.13
>>>   @@ -439,5 +439,10 @@
>>>        public boolean providesCustomization()
>>>        {
>>>            return wrappedPortlet.providesCustomization();
>>>   -    }         +    }   +
>>>   +    public Portlet getPortlet()
>>>   +    {
>>>   +        return wrappedPortlet;
>>>   +    }         }
>> I was reviewing the differences between the security code I have  
>> hanging for Jetspeed 1.3b3 and the current one, and I found this  
>> commit. It completely defeats the facade purpose, which is  
>> authorization.
>> If it is really needed, we could as well completely remove these  
>> classes from the system, or at least move them under a different  
>> package "security" is a misnomer.
> That sounds like an over-reaction.
> Please don't remove the security wrappers. That would break countless  
> applications.
> I hope you are speaking out of frustration.
> Rename the package if it makes you happy, I don't care.

I was not planning to remove them. Yes, possibly I'm overreacting. No, I 
don't think it makes sense to rename the package.

>> I introduced these classes with two purposes:
>> * make clear where the authorization code would be in the system and  
>> ensure that no code out of the core system could fake authorization.
>> * even more important, have opaque facades for the third party  
>> portlets. This helps cleaning contracts between portlets and the  
>> container. It "enforces the Portlet API". If a method is discovered 
>> to  be missing from the scheme, we can add it (both in the base 
>> interface  and here). So, the wrappers would act as "barriers" against 
>> API  pollution.
> They are not a barrier. You can change the security classes just as  
> easily as the API.

Yes, but no "consumer" of the class will be allowed to access public 
methods in subclasses by casting. This was heavily abused, and was 
spoiling the code more and more. In this sense they have acted as barriers.

> As for the Jetspeed Portlet API, do we even want to bother discussing  
> it at this point?

Not tha API itself, but I would like to have Jetspeed code base ready 
for migration (and understand dependencies) for when the API is released 
and discussions begin.

That also is motivating my "clean unused imports" campaign ;-)

>> http://nagoya.apache.org/eyebrowse/ReadMsg?listName=jetspeed- 
>> dev@jakarta.apache.org&msgId=99099
>> was one of the last messages before I stopped tracking HEAD. There I  
>> expressed my concerns about using transparent proxies, since they  
>> defeated the authorization feature. I would rather remove the  
>> Interceptor than having it opened, since it gives a faux feeling of  
>> security.
> You can't just remove the security feature from Jetspeed.
> Paul and I worked very hard on putting in declarative security into  
> Jetspeed, and decoupling Jetspeed security from Turbine.  Programmers  

I know you worked very hard into this. I will not remove anything. Sorry 
if I sounded like I would.

> have complete access to the security API, the full source code. In  
> short they can do whatever they please. Once Jetspeed has a clean  
> separation between container and portal engine, these issues will  
> become more important.
>> Since the method is called with such a common name, I'm not able to  
>> find where it is being used. Could anybody clarify?
> In reviewing the commits, I can't find the cast problem I was having.
> It must have been in some client code.
> I know it was very important, or I wouldn't have committed it.
> My vote is remove the method, and I will have to leave it in for my  
> project.
> It will make extra work for me, but Im used to that when working with  
> Jetspeed.
> I don't have to worry about third party portlet programmers breaking  
> into security in my applications.

I will try to remove it in my copy, and see if something breaks. After 
this, I will warn and commit the change.


> -- 
> David Sean Taylor
> Bluesunrise Software
> david@bluesunrise.com
> +01 707 773-4646
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jetspeed-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jetspeed-dev-help@jakarta.apache.org

To unsubscribe, e-mail: jetspeed-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jetspeed-dev-help@jakarta.apache.org

View raw message