Comment 26 for bug 1842947

On Fri, Dec 13, 2019 at 01:58:19PM -0000, Dan Streetman wrote:
> > I don't see a specific problem it would fix, nor do I see a significant future
> > risk of a problem being exposed.
>
> the [test case] section shows how this problem can manifest; and this
> *actually did happen* with the bionic build, as explained in [impact]
> section.

Because of an attempt at a feature addition, right? And that was picked
up at verification time? I don't think we expect to make any further
feature additions to Xenial now?

> I honestly don't understand the resistance to what seems completely safe
> and correct to me, nor do I understand if my [impact] and [test case]
> sections are unclear, so maybe I'm missing something in @racb's logic.

"We never assume that any change, no matter how obvious, is completely
free of regression risk." --https://wiki.ubuntu.com/StableReleaseUpdates

Changing the build time configuration of a package seems ripe for
unexpected and unrelated regressions to me. Further, dpkg is a critical
package and a regression in it may not be easy for users to resolve just
by us releasing a further update.

I suppose we could stage a change in proposed, because we do now have
that mechanism, but depending on the nature of a future patch, taking
the risk on this change might be unnecessary in the future so I don't
think that would mitigate my concern here.

Your change also seems in principle "completely safe and correct" to me,
but take a look at recent 'regression-update' tagged bugs to see how
other SRUs that seemed "completely safe and correct" resulted in
regressions.

Carefully weighing up the benefits and the risks and reaching a
different conclusion is perfectly fine. It concerns me that rather than
doing this you seem to be denying the existence of any possibility of
regression risk just because it seems "completely safe and correct" at
the moment. That's not how regression risk works.