Re: Formal objection: Changing how the testing of potato works would invalidate the whole test. So please don't change it.
This is a tough call, but I suggest that it may result in a net saving of
time to abandon the current test cycle. We know, for example, that the
kernel is going to have to be changed before release, first because it is
bad on general principles to release on the 2.2.15pre20 kernel instead of
the official 2.2.15 kernel, and second because there is at least one
critical bug (the Adaptec SCSI problem, bug 63946) which is going to force
a kernel patch.
I don't mean to start an argument here about whether there are any
significant differences between 2.2.15pre20 and 2.2.15 as released,
because I consider that irrelevant: the credibility of the whole
distribution would be severely damaged by actually releasing a "pre"
kernel, and I doubt that anyone really intends to do it.
Certainly, the auric move is a one-shot deal that will never be repeated,
so we can look at this as just a run of bad luck. If a new set of boot
floppies can really be ready in a week, then I think it makes most sense
to abandon the current test cycle, apply whatever fixes for
release-critical bugs are present in incoming, and go from there.
If we are talking about changing things on the level of the kernel, and I
think it is fairly definite that we are, then I have to regard it as
doubtful what we would learn from continuing the present test cycle.
On 2000-05-15 at 20:22 +0200, Richard Braakman wrote:
> I think it's good that we tested boot-floppies and CD image generation
> -- obviously it needed the test, since it didn't work. The boot floppies
> are ready now, but we still have no images. I think we lost about a week
> to the auric move, however. I hope that next time it will go faster.
* * *
> We'll also throw away the boot-floppies we have now. So I'll pose this
> to the people working on them: What do you prefer? Go with 2.2.13 for
> a two-week test, or stop this cycle and prepare new ones about a week
> from now?