[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [virgo-dev] Security in Virgo

For the record, we do not recommend the use of "dynamic import *" as it represents unspecified dependencies on arbitrary packages which are not known until class load time. Sorry, but there we are.


On 22 Nov 2010, at 12:41, Ric Klaren wrote:

> Hi,
> On 20 November 2010 01:38, Dmitry Sklyut <dmitry@xxxxxxxxxxx> wrote:
>> Can you comment on acl changes need?
> We used the standard jdbc implementation and ran into I think a class
> loading issue in the
> BasicLookupStrategy.convertCurrentResultIntoObject() on the line that
> a new ObjectIdentity is created (we copied the implementation from
> spring security 3.0.0... but if I look at the code again I'm not
> directly seeing a problem for class loading in the original version,
> so my memory may play tricks on me here).
> Anycase in our integration we split out the creation of the
> ObjectIdentiyImpl to a resolver bundle with a dynamic import of *, so
> only that bundle would be restarting in case of pieces of domain being
> reloaded in stead of the close to the whole stack. Next to that this
> resolver is used in our service for Acl management (in there I do see
> some bits that could give class loading issues (or the need of a
> dynamic import *). So I guess the real need for some glue (the
> resolver) could be caused by how we implemented ACL management. So the
> glue may not be necessary a little bit depending on how you use the
> JdbcMutableAclService.
> Cheers,
> Ric
> _______________________________________________
> virgo-dev mailing list
> virgo-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/virgo-dev