Re: Formal objection: Changing how the testing of potato works would invalidate the whole test. So please don't change it.
On 2000-05-16 at 11:16 +0100, Julian Gilbey wrote:
> On Tue, May 16, 2000 at 05:26:28AM -0400, Mike Bilow wrote:
> > I am not wedded to any particular alternative, and it would be reasonable
> > to choose either course, but the release manager asked for opinions and I
> > gave mine. I agree that we have learned a lot from the present test
> > cycle, but this was not quite what we were expecting to learn...
> Your point was primarily "we can't release potato at the end of this
> cycle because of important RC bugs, so let's give up on the test
> cycle." But you missed the point that there's more to the first test
> cycle than seeing whether the release is ready: primarily are the
> boot-floppies and CD teams ready. RC bugs within individual packages
> are less important for this part of the testing.
Look, at this point I just want to see potato go out the door. I will
support whatever procedure consensus feels will accomplish that goal in
the fastest possible time without a significant compromise in quality.
My issue is that I can see no benefit to spending two weeks from this
point testing something based on a kernel we know we cannot release. The
deciding factor is how soon the next boot floppies can be ready. If that
will take two weeks, then we may as well do a test cycle while we wait for
them. On the other hand, if the boot floppies can be ready in a week,
then I see continuing this test cycle as simply adding a week of delay to
test something we know cannot be released. I have seen what amount to
opposing estimates on this issue from the release manager and the boot
floppies manager, so it would be foolish for me to express an opinion.