Salut,
Hubert Iwaniuk wrote:
> Hi,
> see inline
>
> On Wed, Feb 4, 2009 at 10:19 PM, Jeanfrancois Arcand
> <Jeanfrancois.Arcand_at_sun.com <mailto:Jeanfrancois.Arcand_at_sun.com>> wrote:
>
>
> They do now (correct me if I'm wrong), so it could be improved.
>
>
> Which one? I don't think they are going there IMO.
>
>
> @Justin what did you mean by generated artifacts?
>
>
>
> On Wed, Feb 4, 2009 at 9:19 PM, Justin Lee <Justin.Lee_at_sun.com
> <mailto:Justin.Lee_at_sun.com> <mailto:Justin.Lee_at_sun.com
> <mailto:Justin.Lee_at_sun.com>>> wrote:
>
> Generated artifacts should never go under VCS control. They
> can be
> generated easily enough at build time.
>
>
> Hi,
>
> I would like to review our build configuration.
> What do we what build to offer us, except for building?
>
> 1. [must] Package bundles so OSGi users are happy (have it)
>
>
> Right now I think the OSGi bundles are OK, but I do agree with you
> that we can probably aggregate some of them. The current build
> issues are mostly related to the all-in-one OSGi bundles we are
> generating under
>
> trunk/modules/bundles
>
> The main issues there is the anrun plugin that is quite bougus IMO
> when it is time to aggregate the src and javadoc of the aggregated
> modules. I had a hard time making it working but I do still see
> failure with strange exception. I would think this is the first
> place to look at.
>
>
> I agree.
>
>
>
>
>
> 2. [nice] Hudson should be able to deploy generated artifacts
> from SNAPSHOTS to public repository (have it)
>
>
> Yes it does it right now but fail because of the extremely poor
> java.net <http://java.net> infrastructure. We have a workaround now
> internally and we will move to it so deployment stop being broken.
>
>
>
> 3. [nice] Do we want to have generated java doc stored in trunk
> (now), or have it generated and deployed to site while
> performing release? (could have)
>
>
> I think the release profile is quite nice as javadoc & src are only
> generated at that time.
>
> One thing I would like to propose is as soon as a unit test fail and
> breaks the build, we file a p1, include the test, so we can have
> artifact generated.
>
>
> Like automagically? That would be so awesome, but don't see yet how to
> do it.
The magic here will be the commiter..the first that spot the problem
just execute :-) Like right now it is party time with Hudson (red - red-
green- yellow) time so we need to fix something :-)
A+
-- Jeanfrancois
>
>
>
>
> What do you think?
>
>
> That trunk/modules/bundles first thing to review.
>
> H.
>