Grant Ingersoll wrote:
I'm confused ... why only copy the forrest generated files and not the
javadoc generated files? ... i thought the plan was to rely on real flat
files on lucene.apache.org to keep the URLs nice and clean and protect
ourselves against the possibility of Hudson (or the whole lucene zone)
being down?
http://www.nabble.com/Lucene-nightly-build-failure-tf3489583.html#a9875495
Yes, that is the plan. Now it needs to be implemented.
I don't see that nightly builds need to be on lucene.apache.org.
lucene.zones.apache.org is a fine host for such build artifacts. No one
should rely on nightly builds. These are not releases, and, for legal
reasons, we should not not make them look like releases.
I don't know
about the whole Hudson thing anymore, as I'm not sure it is adding much
value over just running the script. I'm tending to think we should just
revert back to the script and the old way of storing builds, etc.. The
only thing about the script I would like to see changed is it to be
refactored into smaller scripts that can be run/tested separately.
One nice thing that Hudson does is make sure that we're not running
multiple builds at once in the zone, which has caused problems in the
past. Hudson also has lots of bells and whistles that Lucene Java is
not yet using, like: automatically testing patches as they're submitted
and adding a comment to Jira indicating whether they've passed; listing
new Jira issues incorporated in each nightly build; full log output
available over the web to help diagnose failures; etc.
Doug
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
