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

Re: wrong multiarch hint for dolphin-owncloud[src:owncloud-client]

Hi sandro,

On Sun, Sep 10, 2017 at 06:06:56PM +0200, Sandro Knauß wrote:
> so far as I understood multi-arch correctly i can't add Multi-arch: same to 
> dolphin-owncloud, because dolphin-owncloud depends on owncloud-client. And 

I think that's a common misconception. Having dependencies is not
necessarily a reason to not mark a package Multi-Arch: same. There are a
pile of libraries that are marked Multi-Arch: same and depend on
packages not marked. It just means that the annotation is not useful
(yet). There is no requirement that we apply Multi-Arch annotations in a
bottom-up manner.

> owncloud-client a files under /usr/bin/owncloud, so owncloud-client is not 
> multi-arch co-installable and because of this also dolphin-owncloud, can't be 
> Multi-arch: same.

It is correct that owncloud-client is not coinstallable with itself.
That's why it isn't marked Multi-Arch: same. Very likely, it should be
marked Multi-Arch: foreign and then the suggested marking on
dolphin-owncloud can be exercised.

So technically the hint is not wrong. It just happens that applying it
is not going to help anyone immediately.

> It would be nice to have this additional written with the hint and/or lower 
> the priority of the hint, so I don't stumble aboutt it that often.

Lowering the priority of non-exercisable hints looks like a good idea to
me. I don't have a good idea on how to implement this, because
formulating dependency satisfaction problems in sql tends to be
difficult and expensive. Do you see a good way to improve the
explanatory text in the wiki?

Given that the hinter generates ~10000 hints and the number of hints
tends to go up rather than down, I'd rather work on lowering that number
than invest more time in fiddling with hint priority. Surely, something
must be wrong when every fifth package is buggy wrt. Multi-Arch.


Reply to: