First, awesome work to convert to a Gradle build. This will give us a lot more versatility for modern Java builds, including publication onto Maven. Other functionality that is converted over can be presumably done, since Gradle > Ant in terms of functionality (given the ability to run Ant tasks from Gradle).
Next, this list seems comprehensive to me, because it removes from the Jython history anything that is a build artifact, especially large binaries.
The only other thing I can think of is that we may want to revisit lib-python/2.7/ as a module/external link/however this can be done in git :) Right now, we have updated this manually by copying over from CPython, because we have found automatic upstream changes to be highly unstabilizing to our builds, but we should do a better job of syncing with CPython. So maybe this should be modeled as a module for now in git? There may be better approaches, given that submodules might be too independent in their history. But I wouldn't want this specific aspect to hold up our move to GitHub.
Regardless, we don't need perfection. We do want to avoid most unnecessary history that should have never been there because of limitations in how we brought artifacts/code in from external sources; and your list accomplishes this goal.
On Sat, May 20, 2017 at 1:03 PM, Darjus Loktevic <[hidden email]> wrote:
As i just got a basic gradle based build working, thought I'd kick off a discussion about the necessary tasks/plan:
Complete Gradle support (maybe not complete complete, but enough to do a release and regrtest).