Re: RF peer review/help/sponsoring: seqcluster
On 07.08.20 16:17, Andreas Tille wrote:
> On Fri, Aug 07, 2020 at 02:22:27PM +0200, Steffen Möller wrote:
>>> You mention it as "Recommends". Would it make sense to use this package
>>> just with vienna-rna as "Suggests"? In this case the package could go
>>> into main.
>> I frankly don't know enough about seqcluster, yet. Likely so.
>>
>> However, we have it as a build dependency since the vienna-rna
>> associated workflow is tested. And a build-dep on something non-free
>> likely prevents a submission to main, right?
> Sure. But sometimes there are different tests and if we would skip only
> one to get rid of this Build-Depends that could be worth thinking about.
I suggest we leave it in for now, and hence leave it in contrib, and I
talk back to upstream about it.
>> On a sidenote, this
>> dependency on vienna-rna is also what keeps bcbio out of main, which
>> depends on seqcluster and vienna-rna.
> BTW, is it a *workflow* (thus kind of a user application)? If so the
> name python3-seqcluster is *wrong* and should be rather only seqcluster.
> We prefix package that provides a Python3 **module** with python3 and
> not any package that by chance was written in Python3. So please
> consider renaming the package (and the repository) to seqcluster if
> its not a module you include somewhere in some workflow but rather a
> self-standing application.
I had thought about it. From what I have observed, bcbio used the python
API.
The package comes with a 300 bytes executable, so technically, yes, you can
execute something.
I kind of liked that because of pip the seqcluster name is kind of
guaranteed to be unique in the python world. But seqcluster by itself,
this cries for a name conflict sooner or later, like seqlib is troubling
us these days.
What might be nice is to have an additional binary that only includes
the executables, so we are likely to avoid someone coming up with a
seqcluster package in the near future.
>
>>> I need to admit I have not built the package myself since
>>> I was to lazy to add non-free to my pbuilder.
>> Very much understood ;) I just knew I cannot send this to you without
>> having run this in cowbuilder over here, so I had to go for it.
>>
>> I am happy to just have vienna-rna suggested instead of recommended, so
>> I wait for your comment on the non-free build-dep.
> Only if we can strip the test suite to not necessarily need vienna-rna.
I personally rank the integrity from testing higher than having it in main.
> BTW, did you contacted vienna-rna upstream to free their code and if
> yes, please add this attempt to
>
> https://wiki.debian.org/DebianMed/SoftwareLiberation
Yes. Will do.
I was hoping to run into the guys at one of the upcoming conferences but
... Winter is coming.
I'll collect opinions for another while and eventually then upload.
Many thanks!
Steffen
Reply to: