Relevance of Monolog? (was Re: Monolog / Seb)

[Jeff Mesnil <Jean-Frederic.Mesnil@xxxxxxxxxxxx>]

Hi,

Discussions related to log occurs in several ObjectWeb mailing lists more often 
than not (e.g. in JOTM[1], JORAM[2][3], JOnAS[4][5][6], architecture[7],... the 
list is not exhaustive).

It shows that there's a *real* problem with logs in ObjectWeb.

As I undestood it, Monolog was designed and developped at a transition time when 
logs in Java were blurry: there was a great log tool (log4j) with a big Open 
Source community but Sun decided that it was not the good way and implemented 
their own log in Java 1.4 (JavaLog).
It was not clear then which log systems will be used and Monolog was designed to 
fill the gap between *any* log system. So far, so good.

But maybe it's time now to reevaluate ObjectWeb position regarding log.
It seems that log4j is here to stay and that JavaLog doesn't really take off for 
various reasons (some of them are discuted in a thread[8] on TheServerSide).

Moreover, let's face it, Monolog hasn't gained any ground outside ObjectWeb. And 
even ObjectWeb users complain about it: when lot of different people asks us 
about the relevance of Monolog, maybe we should also ask ourselves the question...

To my mind, Monolog has been useful in a transition time but now that it's 
clearer, it seems to me that log4j and commons-logging are the way to go:

o commons-logging[9] is a wrapper above log systems (1 class LogFactory, 1 
interface Log)
o it is *simple to use*[10]
o its principle is *very similar to JAXP* (a standard way to get a XML parser)
o it supports log4j *and* JavaLog and others
o it's part of a *huge Open Source community* (Jakarta/Apache)
o it has a *lot of contribution*
o it is *widely used*

Of course, it doesn't answer all our needs, e.g. configuration file is dependent 
of the log system. But is that such a big deal?
And instead to do the same in our corner (without a lot of success, I've to 
admint...), wouldn't it be more interesting to contribute to it so that it meets 
our requirements?

I'd like to know what is the point of view of the ObjectWeb community about it 
(the college of architects, the developpers and the users).

I'd pleased to use commons-logging (and log4j) in JOTM as a proof of concept.
Given the architecture of JOTM which centralizes log management, it'd painless 
to do so.
And so I'll be able to tell "by the code" that it's answering JOTM requirements 
and, much more important, JOTM *users* requirements.

I think this would be the best way to go for ObjectWeb community...

regards,
jeff

p.s.: I'm in holidays next week (some snowboard fun in the Alpes!) so I wouldn't 
be able to participate to that discussion before next week.

[1] http://www.objectweb.org/wws/arc/jotm/2003-02/msg00085.html
[2] http://www.objectweb.org/wws/arc/joram/2002-09/msg00005.html
[3] http://www.objectweb.org/wws/arc/joram/2002-07/msg00019.html
[4] http://www.objectweb.org/wws/arc/jonas/2003-01/msg00128.html
[5] http://www.objectweb.org/wws/arc/jonas/2002-12/msg00110.html
[6] http://www.objectweb.org/wws/arc/jonas/2002-03/msg00157.html
[7] http://www.objectweb.org/wws/arc/architecture/2002-06/msg00005.html
[8] 
http://www.theserverside.com/home/thread.jsp?thread_id=17946&article_count=27
[9] http://jakarta.apache.org/commons/logging.html
[10] 
http://jakarta.apache.org/commons/logging/api/org/apache/commons/logging/package-summary.html


--
Jeff Mesnil     INRIA - ObjectWeb Consortium
email: jmesnil at inrialpes.fr
homepage: http://sardes.inrialpes.fr/people/jmesnil/perso.html