Hi Russel,

On Tue, Oct 7, 2008 at 8:54 AM, Russel Winder <russ...@concertant.com> wrote:

It's not so much about the fact Maven 1 was dying or because nobody understood Maven, but it's really because the Groovy build really became messy and impossible to understand. It wasn't really so much Maven 1's fault but more ours because we failed to properly rewrite it / refactor it early enough to keep it maintainable. That said, the death of Maven 1 was clearly a good reason to move to something else at that time. So back in the days, the decision was made to change, and we just had to choose something else. Today, the need to change is not there (yet?).

Right, emotions were definitely involved there. But they were involved in the choice of a new solution, once the decision had been taken to change. It's not the same situation as today where nobody complains with the Ant build. So the emotion is more on the side of those two of you who want to force Groovy to use their solutions :-)

Agreed. But I don't hear complaining voices these days on the Ant build.

Our build is complex, so it can't just be 10 lines of build script. But it has the merits to be way more readable and manageable than the Maven mysteries of lifecycle. Here, with Ant, the logic, the dependencies between targets, are much clearer. Can you tell me that the Groovy build will be 10x shorter with Gant or Gradle? What are the arguments that make you think Gant or Gradle are better? That's really what I want to hear. If you fail to find good arguments, I don't see how customers, companies can agree that Gant / Gradle are better than what they currently have. It's really a technical marketing excercise I'd want to see here. It's not just because project X decided to use Gant / Gradle that it must be good, it's because Gant / Gradle are superior because of x, y, z. So?

I'm happy to see a Gradle build to better understand its merits, backed with such proofs as a running build. And hopefully it should really cover all the use cases, all the paths of the current Ant build to really be able to compare.

I think it's an unfair comparison -- apples and oranges? Here, so far, the sole argument I've heard in favour of Gant / Gradle is that it would be good marketing for those projects if Groovy adopted their build system. And also that it could help bring a bit more people to know and use Groovy. Aren't there any other good reasons beyond these??? I don't understand why neither Hans nor you have succeeded telling me some other good and valid reasons. I'm sure there are, and I want to hear them!!!

It's in much better shape than the old Maven 1 build. I'd be interested in knowing what kind of refactoring it'd deserve, and how Gradle would more elegantly solves what you think is not right in the Ant build.

"If a severe refactoring is in order, then a review is in order": a review first can tell us if a severe refactoring is needed. Let's not put the ox before the cart :-)