On Aug 19, 2009, at 11:21 AM, Pete Muir wrote:
I think we are making progress here, and actually there isn't too
much work to do. Let me paraphrase to make sure I am understanding
you correctly.
* The source of messages should be abstracted behind an interface. A
default implementation that uses resource bundles could be provided
for easy ootb experience.
Yes,
* The provision of the Locale should be an orthogonal concept to the
logging of messages and the creation of the Logger. This should be
handled via the MDC.
"could be handled via the MDC". There are other ways to do it but
that is probably the simplest.
IMO there are still a few questions to answer:
* Takeshi original design allowed you to use Enums as keys for
messages. I'm personally not sure about this, I'm not really sure
what benefits it offers over Strings as keys
Enums are usually associated with an integer "value" and a String
description. The only value is that you might be able to get some
optimization by using the integer value of the enum in some cases. I
don't know that it is worth the effort to support it.
* How does a framework provide translated messages [1] to i8ln
layer? Can we also use the MDC here to push them in, and have the
appender read them?
I'm not sure I understand the question. The framework will provide a
method like:
public String getMessage(String key, Object[] params);
However the messages are stored is hidden behind the implementation of
that interface. Something has to call getMessage(). Potential places
are a) the application before calling SLF4J b) SLF4J before calling a
logging implementation, c) In the logging implementation - such as in
the Appender.
* Should there be a convention for the key used? For example a key
should use a FQN.
No. The key is essentially a contract between the application and
whatever the backing repository is. Everything else should be ignorant.
* How does the integrate with the various slf4j back ends? Do we
somehow want to map it to the JDK logging i8ln support? What about
log4j?
If you take the approach I recommend - do it as late as possible -
then it is outside the scope of SLF4J. It might be done in the
Appender for the logging framework in use or it might actually happen
in the application that is reading the log files based on the Locale
of the user reading them.
[1] I'm assuming that a framework developer will want to provide
translations for it's logging messages
No. An I18N framework would only provide the format the messages have
to be stored in. Applications using the framework will provide their
own message sets and presumably have experts who can localize the
message sets provided by the development team into the desired locales.
Regarding your suggestion to enhance logback - I want to be able to
provide l10n for my log messages regardless of which logging back
end is in use (for example, the framework I am currently working on
currently targets JBoss AS, GlassFish, and Tomcat as possible
runtimes, which use a combination of JDK logging and log4j).
Again, that depends. For example, Lillith (http://freshmeat.net/projects/lilith-viewer
) is an event viewer for Logback. It would be very appropriate for it
to do this. Likewise with Apache Chainsaw for Log4j. Doing the L10N
further up the chain limits the flexibility these tools can provide.
Sometimes that is desirable, but my recommendation is to always start
from as far back as possible and move up the stack only as far as you
have to.
Ralph
On 19 Aug 2009, at 17:16, Ralph Goers wrote:
On Aug 19, 2009, at 8:31 AM, Pete Muir wrote:
Hi Ralph,
Whether or not resource bundles suck in our opinion, they are the
standard approach to this so I believe we can't just dismiss them.
Let me rephrase. It isn't just that they suck. In my environment
resource bundles don't work for I18N. I'm not in favor of having
features implemented that I can't use. But my opinion doesn't
really carry any more weight than yours.
I'm also unsure how, in your approach, a framework would provide
i8ln'ized log messages which would be used?
As I've said, I'm not at all in favor of SLF4J "doing" I18N. It is
better to do it under a framework such as Spring's MessageSource
interface where you can either use the default implementation,
which uses ResourceBundles, or easily provide your own. As I said,
I'm also planning on creating a framework to manage
internationalized messages using Commons Configuration and XML files.
In that scenario the application either calls getMessage() and then
passes the resulting String to the log call or the application
passes the message key as the message on the log call, which is the
approach I would suggest. In the second scenario it is up to the
Appender to resolve the key by calling getMessage(). Ideally, the
Appender should be able to tell whether it is getting a key or the
actual message text. This can be done simply by convention - i.e. a
log message like "key=Message1" or by passing a parameter. There is
no real need to enhance SLF4J to be able to do this.
The piece that SLF4J is missing is the ability to detect the target
Locale of the message. But even this can be handled without change
to SLF4J. Just adding the Locale to the MDC would allow the
Appender to do the right thing. Of course, you'd have to use a
logging implementation that supports the MDC.
In short, instead of asking for a change to SLF4J it would make
more sense to me for you to ask for an enhancement to Logback so
that any Appender can have the opportunity to internationalize the
message. Currently you'd have to do that in a Layout, but I don't
believe all Appenders support them.
Ralph
_______________________________________________
dev mailing list
dev@slf4j.org
http://www.slf4j.org/mailman/listinfo/dev
_______________________________________________
dev mailing list
dev@slf4j.org
http://www.slf4j.org/mailman/listinfo/dev
_______________________________________________
dev mailing list
dev@slf4j.org
http://www.slf4j.org/mailman/listinfo/dev
