Richard Lowe wrote: > Danek Duvall wrote: >> On Wed, Jun 13, 2007 at 03:38:52PM -0400, Richard Lowe wrote: >> >>> Except that what Danek packaged, and what I thought Danek had >>> packaged differ, so this doesn't work due to an API change in mercurial. >> >> Oops. What did I do wrong? >> > > I thought you were just bringing the needed fixes back, not taking the > entire package up to after them, so an unrelated change between the two > means we're not passing enough arguments to util.pathto() :) > > If anything, it's my fault for not having tried with precisely those > packages (though I'd still be in the same "What do I do now?" position > if I had).
So this raises the issue Rich and I were just discussing on IRC. What's the correct way to rev Mercurial? Mercurial is (or will be) clearly part of the CBE. Mercurial is in the WOS via SFW. We want Mercurial to be delivered via the WOS. Historically, CBE components are not delivered via the WOS. I personally don't know of an analogous current situation with anything else that's part of the CBE. Mike or Danek? Do you guys have ideas? Do we just establish a "contract" between the CBE & SFW so that Mercurial is only rev'd in SFW after sufficient testing has been done with the CBE? cheers, steve -- stephen lau // stevel at sun.com | 650.786.0845 | http://whacked.net opensolaris // solaris kernel development
