[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#707843: gbirthday: recommends python-evolution which is no longer built



Hi,

On 15/06/13 09:30, foss@rolf.leggewie.biz wrote:
> On Friday, 14 June, 2013 03:26 PM, Emilio Pozuelo Monfort wrote:
>> On 14/06/13 09:22, Emilio Pozuelo Monfort wrote:
>>> ISTR recommending a package not in the archive was a bug, but I can't find a
>>> reference. So if you would like to keep this that's fine with me as it won't
>>> prevent migration of your package. Though perhaps you may want to downgrade it
>>> to a suggest. I leave that decision to you.
>> Found the reference now, see
>>
>> http://www.debian.org/doc/debian-policy/ch-archive.html#s-main
>>
>> "must not require or recommend a package outside of main".
>>
>> You could downgrade it to a suggests though.
> 
> 
> Hello Emilio,
> 
> thank you very much for adding that information.  I am certainly not trying to
> keep that Recommends line around forever even though the package is gone.  My
> intention is to fully understand what you and Jeremy are trying to fix and to
> make sure it is fixed in the right way.  In particular I was tripped by the
> suggested change not being to simply drop the recommends line but making changes
> to the build where the added patch 03_disable_evo_database.patch claims in the
> subject that the evolution backend no longer works.  All I can say is that the
> last time I tried it (admittedly on Ubuntu precise) it was working just fine
> IIRC.  This might no longer be the case with the latest Debian and Ubuntu
> releases which I only run in virtual containers on an as-needed basis.

I don't know what Jeremy did, but I never suggested to touch the build system or
anything similar. From my initial report:

"Your package recommends python-evolution, which is gone. Please remove that
recommendation from your package."

The only problem is with the recommendation. You can drop it or demote it to a
suggests and everything will be fine then.

> As far as the quoted policy is concerned, I believe that mostly concerns the
> DFSG.  A package in main should not recommend a package in non-free, for
> example.  Besides, python-evolution IS in main, just simply not in unstable. 
> So, it is my understanding that the gbirthday package in its current form does
> not violate the policy.

It is about that, but more importantly it is about having a self-contained
suite. E.g. if a package in unstable depends on something that is not in
unstable, then it *can't* be installed and violates Policy per the above.
Similarly if a package in unstable recommends something that is not in unstable,
it violates Policy as well. I think it is very clear.

Emilio


Reply to: