Carla Mott wrote:
>
> Hi all,
>
> I just posted the one pager for v3 logging and posted a request for
> review on the forum. Please provide comments on that thread by Jan 30
> 2009.
>
> http://forums.java.net/jive/thread.jspa?threadID=56532
Hi Carla. Some comments on your spec...
3.1 - Why is syslog only supported on Solaris? Don't we have a pure Java
syslog client API that can log to syslog on any machine?
4.1 - Why did you move logging configuration out of domain.xml and into a
separate properties file?
4.1 - Are there AMX APIs for manipulating logging configuration? Will they
continue to work in v3?
4.1 - Does the logging service need to be started first to make sure logging
is available before anything that might want to do logging?
4.1 - Do you add a shutdown hook to make sure the log is flushed?
4.5.1 - It's not clear to me what class these methods are on. Does this
class also exist in v2 or is it all new to v3?
4.5.1 - Instead of all the set/get methods, why not just a method that
returns a live Properties or Map object? Also, why Hashtable instead of
Map<Object,Object>?
4.5.3 - Why is FileandSyslogHandler improperly capitalized? Why one
handler for both file and syslog?
4.5.3 - I don't understand what log-filter is, but I guess I need to read
the v2 docs...
4.5.3 - Why do the new in v3 properties have short names instead of
com.sun.enterprise.server names? Why are they underscore separated
instead of dash separated?
4.5.3 - I don't understand rotate-at-time. What is the value? A time
of day? Expressed how?
4.5.3 - I don't understand flush_frequency. Every time you log a message
to the log file you're going to log *another* message to the log file?
That can't be it. How does this relate to max_queue_size?
4.6 - Still references Prelude.
4.7 - Why are the properties camel case here? (FlushFrequency vs.
flush_frequency)