Yep, there is a comment in the code, stating that is the expected condition:

//SLF4J not available or not initialized properly, try remaining possibilities

That it still continues to work, even when they've misconfigured their
end, means it is resilient to failure as a good security framework
should be.

SLF4J maintains the same philosophy - it will _never_ throw an
exception if you've done something wrong or something internally goes
wrong.  Only the classloader will complain if you've forgot a binding.

On Tue, Jul 15, 2008 at 4:34 PM, Niklas Gustavsson <[EMAIL PROTECTED]> wrote:
> On Tue, Jul 15, 2008 at 4:07 PM, Les Hazlewood <[EMAIL PROTECTED]> wrote:
>> I can even simplify the solution even more by removing classes.  I
>> could get this solution to work with 1 Log interface, 1 LogFactory
>> abstract class, and 3 Log implementations - Slf4JLog, JdkLog, NoOpLog,
>> all methods 1 line each - easy peasy.
>>
>> That would be it - 5 files total.  That would be as simple as it could
>> get, again with no class loading issues.  Wouldn't this be acceptable?
>
> Having some more time than for my last reply I went looking at the
> current version in SVN. The current implementation will eat the
> exception that SLF4J throws when only the API is present and not an
> implementation. I'm guessing that's not the behavior you want since
> the user in that case has gone through the trouble to set up the API
> JAR but somehow missed out on the implementation. In that case,
> JSecurityLogFactory would now silently ignore SLF4J completely and go
> for JUL.
>
> /niklas
>

Reply via email to