On 11/28/06, Simon Nash <[EMAIL PROTECTED]> wrote:
I'm responding to Jeremy's comments here rather than in TUSCANY-949 because I believe it's considered good practice to have discussions on the mailing list rather than in updates to JIRAs. Jeremy Boynes (JIRA) wrote: > [ http://issues.apache.org/jira/browse/TUSCANY-949?page=comments#action_12454069] > > Jeremy Boynes commented on TUSCANY-949: > --------------------------------------- > > I don't think we should use profiles in this way. We already use profiles > (e.g. in SDO) to determine the type of content that gets built (e.g. in SDO > we use them to generate aggregated vs. non-aggregated javadoc, in SCA we > have a sourcecheck profile that enables checkstyle and pmd) but this is > using them to determine which modules get built. I don't think this allows > us to do something like run sourcecheck on the release modules for example. > The -P option of mvn can have multiple comma-separated values, so there's no problem with specifying -Psourcecheck,release to sourcecheck the release modules. > This is really just trying to get modularity without doing the modularity > work. It would be better to do that by restructuring the build into an > appropriately modular tree. > This patch is fixing the problem of publishing M2 artifacts that should not be published to maven, as discussed on yesterday's IRC chat. It does not prevent us from doing a build tree restructure for better modularity. If at some future time this code becomes redundant because of the restructured tree, it can easily be removed.
Looks like that answers all the questions and sounds convincing to me. We discussed doing this the other day and agreed it needed doing and based on that Simon went ahead and did the work for it, so I think we should go ahead and apply this to M2 now. Unless there are any other concerns I'll do this tonight. ...ant
