[Debian-med-packaging] metastudent_1.0.7-1_amd64.changes REJECTED

[Laszlo Kajan]

Dear FTP Masters, Med Team, Charles, Andreas!

On 21/04/13 00:30, Charles Plessy wrote:
[...]
> my point of view is also that if a package does not contain non-Free files,
> then we should accept it in main.  In the case of metastudent and
> predictprotein, the packaged tools would not be able to function without the
> BLAST databases, and as a consequence it would not be possible for another
> package to depend on metastudent or predictprotein in order to run their tools.
> In that case, I think that it would be important to mention in the descriptions
> of the packages that the users would need to build or download large BLAST
> databases.

Indeed, I will write this into the package descriptions.

> Conversely, a helper package that downloads the BLAST databases as part of its
> installation would belong to contrib, as it can not work on a system or network
> where only Debian is available (no Internet).
> 
> In parallel to this, it would also be interesting to provide the scripts or
> commands that are used to produce the BLAST databasess using reference data.

Yes. In the case of predictprotein, we have made the effort. I will see if the description on the man page can be improved, and all scripts
included.

@FTP Masters, Luca: what solution do you see?:

1: Provide small dummy data inside package, so that the package will work without downloading big data files, but results will still be
practically useless (this would only circumvent the policy issue, not the real problem of the large data requirement, so I have a bad feeling
about this).

2: Accept to main, with a clear message in the long description about the need to download large data files for the package to function (my
preferred solution).

3: Put package in contrib.

4: Ask some wider forum (which one? d-devel?).

Thank you for helping me resolve my big data dependency problem.

Best regards,
Laszlo