Re: Status of python-biopython packaging (Was: Adding python3-biopython)

[Philipp Benner <philipp@debian.org>]

Hi Andreas,

> Moreover we should create a separate python-biopathon-tests package
> containing all the tests and the needed data to create an autopkgtest.

This sounds like a good idea. As far as I understand it, with an
additional autopkgtest we wouldn't have to build depend on all the
packages required for the tests, which I personally would favor.

> Finally we should discuss the role of Depends / Recommends / Suggests.
> I have adjusted the Build-Depends to be able running all tests with
> the tools we have in main.  IMHO it makes sense to set all these tools
> in Recommends.  We could also suggest the non-free tools (paml and
> embassy-phylip).  I fail to understand your motivation to suggest only
> a subset of the tools which are needed to run the test and why you are
> only suggesting these instead of recommending.

For a package to be recommended the user should usually want to install
it together with biopython. I believe that this is not true in this
case. For instance, I'm using biopython mostly because it provides me
with basic data structures which are fully contained in the package
itself. So there is no need to install any further packages. Another
example is a lab that works on phylogenetic reconstruction. In addition
to biopython, the users might want to install PAML as well, but might
have little use for any other package that biopython provides an
interface for. So my guess is that most users do not require all of the
additional packages, which is why I believe that suggesting them is the
better choice. But I agree that the list is outdated and should include
also packages such as PAML and embassy-phylip.

Best,
Philipp