On 05/06/2007, at 5:25 AM, Jochen Wiedmann wrote:
Hi, Brett,
On 6/4/07, Brett Porter <[EMAIL PROTECTED]> wrote:
The fix for MCLIRR-2 is not really correct. This makes the @execute
phase irrelevant, and it causes clirr:clirr to fail on something like
maven-model where the sources are generated.
I think the problem is that executedProject doesn't appear to be
working under Maven 2.0.6 for reports in the site.
I won't refuse your changes, but I'd like to understand.
1.) Why do you think that executedProject should be used? When I
added some
debugging code, I found that executedProject to contain real
nonsence.
(An empty project, or something like that.)
Do so on maven-model (where it contains generated sources) - this
will demonstrate the issue.
2.) Why do you thunk that the existence or nonexistence of
compileSourceRoots isn't
relevant?
Clirr works based on the existence of classes, not sources (it's
doing binary compatibility checking), so it's actually what ends up
in target/classes that's important.
The solution isn't perfect (eg, a resources only jar would still get
unnecessarily checked), but it's more accurate.
Hope that makes sense :)
- Brett
---------------------------------------------------------------------
To unsubscribe from this list please visit:
http://xircles.codehaus.org/manage_email
