Hello Daniel,
Curt Arnold wrote:
On Jun 29, 2006, at 11:03 AM, Daniel Carvalho wrote:
Hi.
i would like to have log4j configured so that each logger is appended
to a different file.
The filename would be based on logger's name, with dots replaced by
file-separator. For instance, the logger named "aplic.dada.x1" would
go to
"/MYLOGDIR/aplic/dada/x1.log"
It would be nice to have an appender "FileTreeAppender" with this
funcionality. Then i would configure the root logger to use this
appender.
Acorddingly to log4j rules, all other loggers would inherit this
appender.
But this wouldn't work, because all loggers would use the same
instance of FileTreeAppender, but i want their outputs to go to
different files. One
solution would be the "FileTreeAppender.doAppend" method choose the
file based in LoggingEvent.getLoggerName(), but this would be very slow.
All you would be adding is a get() on a hash map to map logger name
with to get an OutputStream for each call to doAppend. The cost of
mapping logger name to stream will be trivial compared to actual file
io. You would end up having an open OutputStream around for every
encountered log name until the end of the appenders lifetime, but I
don't know if you had a chance to work around that in any other approach.
There could be a workaround: The information about the logger name is
part of each log entry. You could easily write a script which splits
up one big log file into several others at your choice.
Why do you need that? You cannot see the flow of the application after
splitting the log.
This is the wrong list for this discussion. It belongs to log4-users.
Regards
Boris
