Re: [Tails-dev] When should we fix regressions?

On Sat, Sep 01, 2012 at 10:59:04AM +0000, Ague Mill wrote:
> On Sat, Sep 01, 2012 at 12:35:27PM +0200, bertagaz@??? wrote:
> > > I still would like to see this reach 0.13~rc2.
> >
> > Given we're supposed to be freezed, I'm not sure this is a good idea,
> > unless there is has been tested a lot before being merged, and there is a
> > strong commitment to test this in the 0.13~rc2. Is the goodness of this
> > patch worth the effort or risk to include this so lately in the release
> > process?
>
> Ok. Maybe this should had been discussed some more. The process I know
> the best about time-based releases is the Linux kernel. What happens
> there is that there is a window of time where new features get merged
> in. When this window is closed, a first release candidate is made.
> Then, what goes in are fixes for bugs and regressions. Then a second
> release candidate is made, followed by more bug fixes. Repeat until
> kernel is deemed stable enough for a release.

Sure we should have discussed about this before.

I understand the Linux kernel comparison, but this release cycle is a bit
different from the one we made. We for example didn't really planned to do
a RC3. I recognize it doesn't mean it won't happen. But our release
schedule is tight.

Also I guess that the bug/regressions fixes that get into the Linux code
after the feature freeze are more related to this features than old
regressions/bugs that for most of them probably did get fixed before the
feature freeze.

> This delay on boot is a regression (which appeared in 0.12, IIRC).
> I don't see why it should not be fixed as soon as possible. There is a
> second release candidate planed. If it appears to cause problem, then it
> can be reverted before the final image.

Well, it also depends if we'll be able to see if it causes problems.
That's why I was highlighting the need for careful tests of this patch and
asking for commitment on this. It's a regression, but quite a minor one, so
that's why I was wondering if including it was worse the risks at this
stage. I would have preferred to see this included before RC1, given it
was a known regression not that related to the 0.13 testing process.

But I guess you already did test it. If you feel comfortable enough, I
won't block, also because it seems that the first round of tests have
proven that this release doesn't contains so many problems (not to say
none at the moment), so we might be able to take time to test this.

But still, we would have to be careful and extensively test it.

bert.