[slf4j-dev] SLF4J Implementation for Simple Log

Graham Lea grlea at dev.java.net
Sat Aug 13 09:54:18 CEST 2005


Ceki Gülcü wrote:

> For users who do not need more complex schemes, Simple-Log can be 
> sufficient. However, for those wanting to do more complex things, 
> Marker offers significant new power.

Hi Ceki.
Thanks for the response.

I think decisions like this have to come down to the philosophy of the 
library, which for a bridging API is always "lowest common denominator" 
or "greatest possible flexibility" (or somewhere in-between)?
Neither is wrong, they're just different approaches which will cater to 
different needs (and prejudices).

 From my last mail it's probably obvious that I prefer the lowest common 
denominator approach.
As the project leader (which I assume you are), you need to pick the 
direction and be sure a majority of your committers agree with you.
It seems like you're pretty interested in going the flexibility route, 
which is fine by me.
I think between commons-logging and Log Bridge (and probably others I 
don't know about) there's already enough lowest common denominator 
implementations to pick from, so in that respect it's good that you're 
going to be different in a significant way.

 > Simple-Log can delegate calls to methods with a Marker object to the 
appropriate method without the Marker. This is what NOP, Simple, JDK14 
and NLOG4J implementations do.

Seeing as you're expecting very few packages to actually support Marker 
(none to begin with?), I would suggest that the request for another 
level above Debug still be implemented, seeing as this will give 
increased flexibility to users (possibly in the majority) using a 
logging package that doesn't support Markers.
I didn't read the whole 'Trace Level' thread, so you may have already 
laid out reasons why you don't want to do this.
If so, don't bother going into it again, I just thought I'd throw in my 
"two bob's worth". :-)

Regards,

Graham.




More information about the slf4j-dev mailing list