On 18 December 2012 16:10, Chris Muller <asqueaker at gmail.com> wrote:
>>> Also, should I even bother to test this image if you are planning to deploy yet another one without the "RC" in the name?
>>
>> Yes. That's the whole point of having release candidates.
>
> But then the actual release image does not get tested!
>
> It can't be a true candidate unless it's THE one that will be
> released.  We will not release something named RC3.  We should name it
> for real now.

I don't understand this. It's an image that contains the current best
guess at what will finally be released. It may well contain things
that aren't right, so MUST be tested by people. I _expect_ things to
change. Who cares what it's named, other than that it's named
differently to any other RC?

In particular, I _cannot_ version the artifact "correctly" at the
moment. I have seen no stamp of approval from anyone regarding my
ReleaseBuilder and Morphic changes, so my release process moves the
latest successful build (which has been tested on the Linux platform)
past trunk. If people are happy with what I've done so far, let's move
the ReleaseBuilder/Morphic inbox submissions into trunk and I'll
continue hacking directly on trunk.

I see no need to touch Morphic again (unless we want to strip out the
mention of unloadAllKnownPackages because of the current GetText
unloading issues), and changes to ReleaseBuilder won't affect other
users. So if we push these to trunk, I'll happily start producing
properly versioned artifacts. (Ideally, Jenkins would do this
automatically on an all-green build, but we have neither the script
nor the green light to do this.)

frank