[SNMP4J] SNMP4J v1.2RC

[Frank Fock]

Hi Gregor,

Thanks for your tip. I have read about the ClassLoader problems with JCL
but thought that those problems have been solved through the use of
WeakHashMap in version 1.0.4. This must have been a misunderstanding,
since it seems that v1.0.5 will use WeakHashMap for the first time.

The UGLI approach seems to be cleaner. Does anybody have a strong
opinion against changing ot UGLI? I would make to change to v1.2 and
following.

Best regards,
Frank


Gregor B. Rosenauer wrote:

> A good alternative might be Apache UGLI in conjunction with log4j1.3:
>
> http://logging.apache.org/log4j/docs/ugli.html
>
> It does not suffer from JCL's ClassLoader-problems and might save you 
> from hard to find bugs obscured by JCL.
>
> The big advantage is that you can use log4j 1.3's exceptionally *cool* 
> parameterized log messages (see link) instead of being limited to the 
> commons-mini-API.
>
> A minor drawback is that it's only available with log4j *1.3* which is 
> still alpha (though alpha6 is already widely used and stable).
>
> Quote:
> "Authors of widely-distributed components and libraries may code 
> against the UGLI interface in order to avoid imposing an logging API 
> implementation on the end-user. At deployment time, the end-user may 
> choose the desired logging API implementation by inserting the 
> corresponding jar file in her classpath. This stupid, simple but 
> robust approach avoids many of the painful bugs associated with 
> dynamic discovery processes. Life is too short to be spending it 
> hunting down class loader problems."
>
> I'm not biased against one or the other framework, just researching 
> this topic currently and want to share the info I got so far.
>
> Gregor
> _______________________________________________
> SNMP4J mailing list
> SNMP4J at agentpp.org
> http://lists.agentpp.org/mailman/listinfo/snmp4j
>