Thank you for your response. I am grateful that you are willing to
look
into Ceki's concerns. I believe it would be possible to integrate
many, if
not all, of Ceki's issues into the logging API without too much
difficulty.
I understand these things aren't "quick fix" in nature, but far easier
to
change things now than later, when the API is in widespread use.
Of
course, and I admit I am biased, I believe the best solution would be
to
make log4j the standard logging API. It is fully developed, tested,
and
ready to roll. If Sun announced log4j as the standard API, I
firmly
believe the developer community would herald the decision. The only
people
the decision may cross would be those who spent their time and effort
forming JSR47, at which point it becomes a political issue. Even so,
I
understand if it is not possible to replace JSR 47 with log4j and would
be
willing to settle with integrating Ceki's suggestions into JSR47.
Again, thank you for your desire to consider this further. If there
is any
way I can be of service in this area, just let me know.
- Andy
Graham
Hamilton
To:
AndyDePue@kyrus.com
<kgh@eng.sun.
cc:
com>
Groups: Don't
Expand
Subject: Re: Request to use log4j as Java logging
standard instead of JSR 47.
06/13/01
04:35
PM
Andy,
Thanks for your interest in the logging APIs.
As Ceki noted, IBM has been a very active participant in the
JSR-047 logging JSR. When the logging JSR was initiated in
January 2000, IBM had both the log4j work at IBM research and
the JLog work at IBM Tivoli. The Tivoli work built on IBM's
earlier product logging experience and included features like
internationalization that were then not present in log4j.
IBM's involvement in JSR-047 was focused on their JLog
experience which had seen a lot of use in IBM products.
However, Ceki also provided useful input from log4j, and that
has influenced some parts of the logging APIs.
In designing the APIs we've tried to draw from numerous
sources and we've tried to support a wide set of requirements.
It's inevitable that when there is a new standardization
effort it will make some different choices from some of
the existing APIs. We went through a similar experience
with JDBC. It built on a bunch of earlier work, but it
wasn't exactly like any of them. Unfortunately we delayed
adding a JDK logging API for too long, and I think it
is important that we resolve that in Merlin.
Unfortunately we are quite late in the Merlin cycle and
we can't really accommodate major API changes at this point.
However, I will be studying Ceki's comments carefully.
thanks
- Graham
AndyDePue@kyrus.com wrote:
>
> Hello,
>
> I have been involved on at least ten projects where
logging has been
> necessary. With this much experience logging (both building
logging
> libraries and using third party libraries), I can tell you that when
I
> discovered log4j, I had never seen anything like it. It is,
simply, the
> best logging package I have ever used. Not just the big things,
but the
> little details are what make it great. After a thorough review
of JSR
47,
> I see it falls short on many of these details. In all honesty,
if JSR47
> became the logging standard, I would never use it as long as log4j
still
> existed in an up-to-date state.
> Please have mercy on us developers and heed this
appeal!
>
> Sincerely,
>
Andy DePue