The interface would be handy in many occasions. Although I say the
default implementation needs to throw a runtime exception if a given
count isn't implemented in that environment. If we make it easy for
users to configure their own implementations (and contribute back)
we'd likely only need to implement a default SessionListener for
servlet environments. Probably useful to model the interface after
some existing implementation, such as Webshere's
(http://publib.boulder.ibm.com/infocenter/wasinfo/v6r0/index.jsp?topic=/com.ibm.websphere.express.doc/info/exp/ae/rprf_datacounter6.html)
though there might be better ones.

Kalle


On Wed, Feb 3, 2010 at 12:45 PM, Les Hazlewood <[email protected]> wrote:
> Bump.  I was thinking about working on this today, but wanted to get
> some feedback before I touch anything.  Any ideas?
>
> On Thu, Jan 21, 2010 at 5:01 PM, Les Hazlewood <[email protected]> wrote:
>> Hi devs (et. al),
>>
>> In trying to address this issue:
>>
>> https://issues.apache.org/jira/browse/SHIRO-117
>>
>> I've realized that just having the QueryableSessionDAO wouldnt' be as
>> useful as it could be.  The SessionManager interface should probably
>> reflect this added functionality too.  So maybe we should probably
>> have methods like:
>>
>> SessionManager.getActiveSessionCount() : long
>> etc.
>>
>> where these methods return -1 if that functionality isn't supported by
>> the underlying SessionDAO.
>>
>> This is the cleanest thing I can think of - the SessionManager
>> interface must remain constant regardless of what the SessionDAO can
>> support.  Some cache APIs have similar methods because they can't
>> guarantee that the underlying implementations support statistics.
>>
>> Any thoughts?
>>
>> - Les
>>
>

Reply via email to