Also, if you drop Log4j into a web container currently it will automatically 
initialize in a Servlet 3 container.  I’ve always been a bit uncomfortable with 
that and it was originally in its own jar, but others preferred it the way it 
currently is.

Ralph

On Apr 12, 2014, at 7:20 AM, Ralph Goers <rgo...@apache.org> wrote:

> Jackson is used to parse JSON configuration files and JMX is currently 
> defaulted to be always enabled.
> 
> Ralph
> 
> On Apr 12, 2014, at 3:24 AM, Łukasz Dywicki <l...@code-house.org> wrote:
> 
>> Ralph,
>> For these who wants a "big” log4j you might provide aggregate JAR just like 
>> astyanax does. I think it’s much easier going this direction than other way 
>> around. Fine grained dependencies are proof of good library design. In slf4j 
>> epoch developers do not depend on any log4j code except test scope where 
>> mostly used is FileAppender. During deployment time it doesn’t make big 
>> difference too because only one place which would be affected is assembly. I 
>> can also imagine that some people could still love spring log4j stuff and 
>> will not need direct log4j support for it.
>> 
>> I can understand that disruptor might be mandatory to provide asynchronous 
>> behaviour but why JMX and Jackson could not be separated?
>> 
>> Best regards,
>> Łukasz Dywicki
>> --
>> l...@code-house.org
>> Twitter: ldywicki
>> Blog: http://dywicki.pl
>> Code-House - http://code-house.org
>> 
>> Wiadomość napisana przez Ralph Goers <ralph.go...@dslextreme.com> w dniu 12 
>> kwi 2014, o godz. 06:26:
>> 
>>> There is a balance that needs to be struck here.  We don’t want to add 
>>> support for OSGi at the cost of forcing non-OSGi users to have to include a 
>>> bunch of Log4j dependencies.  That said, I suggest the other day that we 
>>> might want to split some things that a majority of our users will rarely 
>>> use into other modules, possible in a companion project to the Log4j kernel.
>>> 
>>> We actually have a few dependencies, such as Jackson, the disruptor, and 
>>> JMX that are optional but really can’t be separated from the core.
>>> 
>>> That said, an analysis of what could be separated is warranted.
>>> 
>>> Ralph
>>> 
>>> On Apr 11, 2014, at 3:40 PM, Łukasz Dywicki <l...@code-house.org> wrote:
>>> 
>>>> Hey all,
>>>> Apachecon has been ended so I am following the promise I did - I’ve taken 
>>>> a look on the log4j build to check what’s going on with OSGi stuff. 
>>>> Currently OSGi bundles are split from main log4j-core which I think leads 
>>>> to maintanance trouble also number of dependencies of log4j-core seems to 
>>>> be very high. If we could split the log4j-core into smaller modules 
>>>> (mongodb, jms, jmx, web) then there will be no need to keep log4j-osgi and 
>>>> it’s submodules.
>>>> According to maven guides [1] "Optional dependencies are used when it's 
>>>> not really possible (for whatever reason) to split a project up into 
>>>> sub-modules”, I think it will also make library design much more clear and 
>>>> will avoid making unecessary inner code deps. Seems that big, aggregate 
>>>> modules, are always grow to size which require major work to split them to 
>>>> make both: maintanance and community involvement possible. 
>>>> 
>>>> [1] 
>>>> https://maven.apache.org/guides/introduction/introduction-to-optional-and-excludes-dependencies.html
>>>> 
>>>> Cheers,
>>>> Łukasz Dywicki
>>>> --
>>>> l...@code-house.org
>>>> Twitter: ldywicki
>>>> Blog: http://dywicki.pl
>>>> Code-House - http://code-house.org
>>>> 
>>> 
>> 

Reply via email to