Announce: Database Backed Storage Service

[Tom Zeller]

>> * Wessel, Keith <kwessel at illinois.edu> [2013-10-03 21:03]:
>>> I also discovered, though you probably already know this, that when
>>> using a container-based database connection, aacli.sh obviously
>>> dumps out an ugly error before returning the attributes. Seems like
>>> another good reason for a parallel $IDP_HOME/conf dir for use with
>>> aacli.sh, one with an internal.xml that uses the standard storage
>>> service instead of the database-backed one if one chooses to take
>>> the container-managed route.
>>
>> Unless you run aacli.sh in a container (which even was considered, I
>> think) obviously any kind of container-managed JDBC won't work.
>>
>> I haven't considered making a copy of my conf/ replacing any
>> container-managed JDBC connections with application managed JDBC
>> connections solely for the aacli case. Instead I stopped using
>> aacli.sh :)
>> For someone needing certain Oracle features (like I did for Univie's
>> IdP, accessing DataGuard or RAC databased) and still wanting a working
>> aacli that's a idea well worth considering. (Though immediately
>> concerns about redundant but not identical configiuration come up ;)
>
> For Spring wiring unique to aacli, maybe override the bean present in
> internal.xml via a new xml configuration file passed to aacli.sh, but
> I would need to look to know if this suggestion is reasonable or
> possible.

Oh, looks like back in 2010 Scott added the --springExts parameter to
aacli.sh which can be used to override Spring bean definitions. Just
updated the AACLI wiki page to include that parameter.