On Wed, Jan 14, 2009 at 2:37 PM, Eric Wilhelm <[email protected]>wrote:
> Ideally, I would (when wearing my release manager hat once per week) > have a green button for "push an alpha" and a green button for "push a > release". I can make the buttons, but the information "go for alpha" > needs to come from e.g. trunk having passed the tests on a sufficient > set of perls+platforms, and the information "go for release" would come > from a week or two worth of cpantesters results. > I was thinking more like a monthly "push an alpha" without a lot of thought about it and then periodically, "push a release" when the list has some consensus that it's ready. > Is testing against svn commits too difficult? If so, what would make it > easier? > I think testing trunk periodically is more meaningful than testing every commit. Particularly for those of us who work granularly with SVK, commits often have small, incomplete changes and then are pushed. Or, for instance, all the tag and branch update work you've done probably don't need to be tested. But even a nightly build & test cycle on the trunk would help give a quick "it's broken" check. -- David
