Re: bcbio is on salsa - builds .debs, importable from python, missig seqcluster (at least) and python(2)-toolz

[Steffen Möller <steffen_moeller@gmx.de>]

Hi Liubov,

On 7/9/18 10:16 AM, Liubov Chuprikova wrote:
> Hi,
>
> On Mon, 9 Jul 2018 at 10:27 Andreas Tille <andreas@an3as.eu
> <mailto:andreas@an3as.eu>> wrote:
>
>     Hi,
>
>     Steffen, if you think Liubov might be able to help at getting tests
>     working it is fine for me if she wants to tackle this.
>
>
> I  would try to fix these tests.
This is very helpful indeed, many thanks!
>
>     @Steffen: May be to get Liubov started its helpful if you provide some
>     more verbose information with links. 
>
>
I admit not to really know myself where to start. python-seqcluster,
pythong-gfftools, python-pybedtools, python-cyvcf2 are all not in our
distribution but only on salsa. So, I think, for the moment I would
already be very happy if a second individual would follow my steps. And
then, I see

 a) get packages (back) into the distribution with an RC bug as long as
tests don't work

 b) investigate (likely together with upstream) and fix the failing tests

 c) come up with tutorials to run the biological workflow(s) with Debian

For python-seqcluster I have forked my own v3 variant at
https://github.com/smoe/seqcluster and sent pull requests. From that
source tree I symbolically link to the debian directory of the salsa
checkout. This seems to work rather nicely with "fakeroot ./debian/rules
binary" to get working packages quickly and communicate with upstream. I
think I will adopt that concept for other packages, too.

So, at the moment we are still at some pre-a) stage. As long as we
target unstable, the one or other failing test is not a catastrophe. If
I do not get to c) reasonably quickly then I will not have any
justification to invest more time for b) .

So, many thanks again for all you can do to help,

Steffen