On 2017-06-27 00:10:02, Dominik George wrote:
> Hi,
>
>> I stand corrected: the following command does call sleep.
>>
>> /var/lib/docker/nuke-graph-directory.sh /var/lib/docker
>>
>> I agree this is a bug, but I disagree it is critical, because,
>> technically, it's not the maintainer script that calls "sleep".
>
> Erm… I somehow feel I missed some kind of "policy interpretation
> creativity contest" ;)?
I win! ;)
> If something used in the maintainer script breaks policy, the maintainer
> breaks policy. With your argumentation, I could move a bunch of bad
> commands to this_is_not_a_maintainer_script.sh, source that, and policy
> does not mean anything anymore.
The thing is that it *is* not a maintainer script: it's some upstream
device called from a maintainer script. The difference is minor, but i
do think it matters, in terms of policy. It's not as if the maintainer
*did* move that to a script, as far as I know.
In other word, the script that calls "sleep" is not in debian/ so in my
mind not covered by policy.
> Besides, calling sleep is not the issue here. The issue is the
> maintainer script requires the package to be removed interactively and
> cannot be controlled in an uanttended run.
I do believe the package can be removed non-interactively.
What I think you want is the possibility for users to disable the
removal of data through some configuration (e.g. preseed). I would argue
this is currently not supported but would be a nice addition. In that
case, the script would simply not be called...
A.
--
Religion is like a blind man looking in a black room for a black cat
that isn't there, and finding it.
- Oscar Wilde
