On08.02.10 18:02, Bryan Pendleton wrote:
Just to be clear, you wouldn't mind that a single existing JavaDoc
warning fails the build?
Indeed, that would be fine with me. The sooner we know about a problem,
the sooner we fix it, in my opinion.
FYI, the job is now on an hourly schedule, the most frequent schedule
recommended by the infra team when polling the SVN repos. It will
only be run if changes in the SVN repos are detected.
So, in your hypothetical example, the build would continue to fail until
the JavaDoc warning was fixed. But, it would not fail every hour; it
would
fail every time there is a SVN change, until the warning was fixed.
I'm OK with it, but others should offer their opinions!
Nobody has objected to the proposed change, I'll implement it tomorrow.
I have introduced JavaDoc warnings several times myself, but making sure
there aren't any before committing is pretty simple. However, it
involves running 'ant javadoc' and looking at the JavaDoc output (ant
reports BUILD SUCCESS even if there is a JavaDoc error / warning).
To simply this a little bit, I created an alias to help me out (note the
space before 'error' below):
ant javadoc | tee javadoc.out | egrep 'warning| error|BUILD'
Note that the committer(s) that made the changes causing the JavaDoc
warnings will get an email from Hudson. If you get one, it doesn't
necessarily mean that your change is the culprit, but you should read
the Hudson log to make sure you're in the clear :)
Again, here are two relevant links:
http://hudson.zones.apache.org/hudson/job/Derby-trunk/
http://hudson.zones.apache.org/hudson/job/Derby-trunk/warnings
Regards,
--
Kristian
thanks,
bryan
