Thanks, just checked the commit.
While we are at it, can you please add an option to specify the bufferSize
of the BufferedOutputStream.
The default buffer size is often fine, but sometimes too small for high
throughput apps.
Along the lines of
old:
if(bufferedIO) {
fw = new BufferedOutputStream(fw);
}
new:
if(bufferedIO) {
fw = new BufferedOutputStream(fw, myBufSize);
}
Wolfgang.
>Wolfgang,
>
>Yes, a day or so ago. It's a preliminary commit, in the sense that the
>code still needs to be completed and tested. Ceki
>
>At 16:14 13.08.2001 +0200, you wrote:
> >Has there been any commit related to (optionally) using
> BufferedOutputStream in File|Socket|SyslogAppender?
> >(and I don't mean AsyncAppender)
> >
> >Wolfgang.
> >
> >>Cool. Of course buffering means log4j should be shut down properly when
> >>exiting the JVM so as to flush the buffer. Here it would be highly
> >>useful to be able to depend on the 1.2 shutdown hook.
> >>
> >>Anders
> >>
> >>
> >>Ceki G�lc� wrote:
> >>>
> >>> Anders,
> >>>
> >>> Do you remember the discussion we had about buffering in FileAppender?
> >>> We now have a real world example showing that you were right and I
> was wrong.
> >>> See http://nagoya.apache.org/bugzilla/show_bug.cgi?id=2341
> >>>
> >>> My performance tests were done on my machine with a single user
> logged on, doing \
> >>> just the tests. In real world situation you have a many processes
> accessing the \
> >>> disk. And in that situation buffering is all the more important.
> >>> Fortunately, this one is not hard to fix. Cheers, Ceki
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
