+1

Personally, I think the TRACE feature is dubious at best.  If it was me
writing for myself, I'd say -1, but if it makes the user base happy and
if log4j is truly user-driven, then let's do it.  It's not going to do
much harm for the rest of us who don't need it.  (But I would have to
draw the line at adding FINE, FINER, FINEST as well.)

For people who like having a TRACE level, I'd like to hear how it has
been helpful to them.  I believe it's one of those features that sounds
good when you use if, but is hardly useful in real practice.

Personally, if you structure your classes into logical packages and put
in useful debug information--NOT spurious junk--then with careful level
filtering you'll get the right amount of useful data.

Here are the common problems I see with people using logging:

1.  Gigantic classes or packages, leading to not enough categories
2.  Log lots of data about things we don't care about.  For example,
instead of showing data points X Y Z, we see A B C D E F ... X Y Z.
3.  Data isn't checked properly.  If the data looks wrong, throw an
exception!  Validate all data if it is easy to do so.  Make the
exception explain what data is wrong, throw exceptions with messages.
4.  Log statements with no reason for existence.
5.  Log statements within large loops that write out 100+ times.

So, for people who can't use logging properly, TRACE isn't going to
improve their situation:  Likely people will feel free to *add more*
logging, which is going to decrease the S/N ratio.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to