Kristian Waagan wrote:
Kristian Waagan wrote:
Rick Hillegas wrote:
Hi Kristian,
I have lost track of where this issue is. Are you happy with
solution (1) or do you still need a solaris zone on an Apache
machine? If you still need a zone, can you give me some
justification--I think that I will be asked why (1) isn't
sufficient. It's ok to say "I don't know, I'm running some
experiments, I'll get back to you." I just want to make sure that
I'm not letting this issue fall through the cracks.
Hi Rick,
The reason I haven't replied yet, is that I don't know how well
suited Hudson (or one of the other CI tools) is for building and
publishing the manuals.
If I get feedback on how the build/publish process of the manuals are
carried out (specifically if it follows the process outlined on the
Derby web page), I may be able to pursue solution (1).
I suggest you ignore my request for a separate Solaris zone for a
little while. I just requested a Hudson account for Derby
(https://issues.apache.org/jira/browse/INFRA-2090).
I have now gotten the Hudson account, and I have created a job
building Derby:
http://hudson.zones.apache.org/hudson/job/Derby-trunk/
It is polling the svn repos (trunk only) every 15 minutes. Since I
just created it, I'm the only one getting the mails so far. I'd like
the community to consider the following two questions:
- Should the notification mails go to derby-dev?
My take on this is yes. A mail will be sent when a build breaks, and
when a series of broken builds start working again.
Yes, please. People who don't want to see these mails can configure
their mail filters. I assume that everyone in a code community cares
about the health of the codeline.
- Should a separate mail be sent to the committer(s) breaking the build?
This will send a mail to all committers having committed a change in
the 15 minutes window.
Yes, please.
Thanks, Kristian.
It should be trivial to start building 10.5 as well, but it may be a
tad harder for the older versions where we don't have the automatic
Java compiler configuration.
I'll think some more about building the docs through Hudson. I guess
building it is simple enough, but the problem Andrew described has to
be solved somehow - how do we get the docs to the web server in a
secure manner?
