David Crossley wrote:
Ross Gardler wrote:
Gavin wrote:
Just wondering,
Would Hudson [1] and its variety of Plugins [2] be of any use to Forrest?
I'm not at all sure. Most of the utility of Hudson is in managing
executble code. We have some code, but it is mostly XML and XSLT. What
little code we have has almost no tests - thus making the continuous
integration efforts rather pointless.
Using such tools might encourage tests to be contributed
and general code improvement to happen.
We do have six modules in main/java/org/apache/forrest/
and some of those are extremely important to us.
We have eight plugins that have some Java code.
All true.
Finally, we don't really have a
great deal of code that depends on other code.
Hmmm ... Cocoon, Avalon, Excalibur, etc. classes are
mentioned a lot in the above code.
Let me clarify. We don't depend on code to the extent that we are
pouching the boundaries of that code and therefore need CI. We run
packaged and released code. All our dependencies are inside those releases.
Having Hudson won't harm us. I'm just not sure well see much benefit.
Having said that, it is true that I get a kick out of seeing the graphs
of tints go up and defects go down. It does indeed encourage good
behaviour. Problem is we don't get all that by installing Hudson, we
need much more than that. If someone is going to put that effort in
then, of course, I'd be happy to see it. I just think there are other
more important things to focus on.
Ross
