[Debian-med-packaging] Bug#503367: plink: file conflict with putty-tools

[Andreas Tille]

On Wed, 29 Oct 2008, Charles Plessy wrote:

> I think that I would like the Debian Blend distributions (formerly called CDDs)
> to manage this smartly in the future. We could have some mechanisms that make
> sure that for biologists, plink relates to SNPs, not to SSH. But this is a long
> term goal with no implementation plan.

Well, an implementation plan could be to symlink any binary to a
     /usr/share/blends/<blendname>/bin
directory and adjust PATH for those users who are registered as
user of this blend - so I see no problem in principle to realise
this idea.  But I'm absolutely not happy about such kind of workarounds.
As I explained earlier it is also about name space polution in the
Free Software namespace - it makes no sense to find a Debian specific
or a Blend specific solution and should be avoided in general.

> If we are confident that the user sets of
> plink and putty are mutually exclusive,

We can and should not be confident about this - neither in this specific case
nor in general.

> I would not mind a Conflict even if it
> is not allowed by the Policy. But how confident are we that we will not get
> complains? We will be in a much worse situation if we have to make changes
> after our userbase is established.

Yes - that's why we should also care for users which are not yet users of
Debian - finally they will all use Debian once we reached world domination. ;-)

> I personnaly would advocate d) because it can be the basis for a more global
> solution later. For instance something like /usr/lib/debian-med/plink ->
> /usr/lib/plink/plink, and populating /usr/lib/debian-med/ with our other cases
> of namespace pollutors-polluted programs. We could then ask our users to put
> /usr/lib/debian-med/ in their paths, so that they do not have to micromanage
> such issues.

Something like this as I mentioned above.  The per package /usr/lib
solution is just established and accepted but only a Debian specific
workaround.

> Lastly, Upstream was very responsive; we probably should discuss again with him.

Pointing upstream to the mails collected inside the bug page might be
reasonable to understand all arguments.

Kind regards

             Andreas.

-- 
http://fam-tille.de