Bwolen Yang wrote:
done.
https://issues.apache.org/jira/browse/HADOOP-1489
Thanks!
For extra-credit, attach a unit test that demonstrates it and a patch
that fixes it!
http://wiki.apache.org/lucene-hadoop/HowToContribute
a couple questions.
- When including code/stack trace, my input was nicely indented, but
after submission, somehow the text formatting fields are all mess up.
(e.g., the entire code segment became one line). Is attachment the
only way to do this right?
You can edit the issue and use wiki markup. There are preview and help
buttons beside the text box. {code}foo{code} or {noformat}foo{noformat}
are probably what you want.
- To track this, I essentially was doing "printf" debugging. It is
rather slow/painful.
That's actually my preferred method of debugging!
I generally find the ability to debugger-attach to programs and just
roam around the stack helps greatly in understanding complex programs
quickly (even if I am not looking for bugs).
If you run with LocalRunner (instead of JobTracker and TaskTracker) then
normal debuggers should work well. But, when things are distributed,
debugging is more difficult.
Doug
