[elbrus at debian.org: Bug#905117: please make autopkgtest smarter than just /bin/true]

[Bas Couwenberg]

Hi Ruben,

On 2018-07-31 14:18, Ruben Undheim wrote:
> Thanks for doing some maintenance on the fyba package, moving to team
> maintenance etc.
> 
> But shortly afterwards now, this forwarded email popped into my
> mailbox. :D And I noticed that you added this (now) illegal
> autopkgtest now when you uploaded just a few days ago. :)
> 
> Do you have any plans to fix this, or should I do it? I have started
> adding some CI to some of the other packages, so it would be a natural
> step for me to continue with fyba.

I just resolved the bug by removing the autopkgtest.

If you're willing to spend time to develop a more in-depth autopkgtest, 
please do.

There is value to autopkgtest, but I don't have time to develop 
non-trivial autopkgtests.

There are simply too many packages for which I'm the only active 
contributor.

It also doesn't help that sbuild breaks every time the kernel is updated 
in unstable and the system hasn't been booted with it yet, thanks to its 
aufs dependency only building built for the latest kernel package.

> Best regards
> Ruben

Kind Regards,

Bas

> ----- Forwarded message from Paul Gevers <elbrus at debian.org> -----
> 
> Date: Tue, 31 Jul 2018 13:03:15 +0200
> From: Paul Gevers <elbrus at debian.org>
> To: Debian Bug Tracking System <submit at bugs.debian.org>
> Subject: Bug#905117: please make autopkgtest smarter than just 
> /bin/true
> User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
> Thunderbird/52.9.1
> 
> Source: fyba
> Version: 4.1.1-4
> 
> Dear maintainer,
> 
> I want to draw your attention to the following quote [1] from the
> release team:
> '''
> For this proposal to make sense, all deployed autopkgtests must
> actually test the package involved to some extent.  We trust it will
> not be necessary to establish a technical solution for this part.
> '''
> 
> I have verified with the release team and they stand by that quote and
> they consider this kind of installability testing via autopkgtest not
> "testing the package involved to some extent" if it is the _only_ test.
> Also, installability is better tested with piuparts [2] (which, on
> failure, will fully block migration). The reason why this is 
> undesirable
> is that packages with a passing autopkgtest (and no regressions
> otherwise) receive a reduced migration age. The release team and I
> consider installability not enough reason to grant that reduction.
> 
> Please enhance or drop the autopkgtest of your package.
> 
> For now, I have restored the original age for the latest upload of your
> package, which will be visible on the excuses page soon.
> 
> Paul
> 
> [1] 
> https://lists.debian.org/debian-devel-announce/2013/08/msg00006.html
> [2] https://piuparts.debian.org/
> 
> 
> 
> 
> 
> ----- End forwarded message -----