[pkg-go] Bug#989145: Please do not use uml fakemachine backend in the DEP-8 test

Lucas Kanashiro kanashiro at ubuntu.com
Mon Oct 24 12:42:14 BST 2022

On 24/10/2022 08:37, Christopher Obbard wrote:
> On Mon, 2022-10-24 at 07:38 -0300, Lucas Kanashiro wrote:
>> Hi Chris,
>> On 24/10/2022 06:32, Christopher Obbard wrote:
>>> Hi Lucas & Steve,
>>> On Wed, 25 May 2022 17:25:29 -0300 Lucas Kanashiro
>>> <kanashiro at ubuntu.com> wrote:
>>>> On Wed, 29 Sep 2021 15:21:01 -0700 Steve Langasek
>>>> <steve.langasek at canonical.com> wrote:
>>>>> My current suggestion is:
>>>>> - ask for an additional autopkgtest in Debian that runs with
>>>>> debos
>>>>> --disable-fakemachine (instead of replacing the current
>>> autopkgtest)
>>>>> - introduce an Ubuntu delta that drops the autopkgtest which
>>> depends on
>>>>> fakemachine
>>>>> - optionally, patch debos in Ubuntu to use --disable-
>>>>> fakemachine by
>>>> default?
>>>> Since we got no reply so far I patched debos to use
>>>> --disable-fakemachine and adjusted the test dependencies and
>>>> restrictions in Ubuntu.
>>>> Hopefully, we'll sort this out in Debian and then we can drop the
>>> delta
>>>> added.
>>> I plan on implementing the following when releasing a new version:
>>> 1) modify the build-chroot autopkgtest to use --disable-fakemachine
>>> in
>>> the debos call and remove the additional dependencies (e.g. just
>>> running on the host, this will need Restrictions: needs-sudo)
>> I think you meant "needs-root" restriction here, right?
> Well, either works for me, but typically a user who is using debos with
> --disable-fakemachine will use sudo to elevate privileges rather than
> running as the root account. This will make the test better reflect
> reality. Does it make sense to use needs-sudo in this case?

Aha, OK. I thought you got confused with the restriction name but no, I 
was not aware of this new restriction, sorry :)

Looking at its description, it seems what you want indeed.

Lucas Kanashiro

More information about the Pkg-go-maintainers mailing list