Re: BioConductor MOFA2 needs basilisk - which installs a conda environment

[Nilesh Patra <nilesh@nileshpatra.info>]

On 9/15/21 2:45 PM, Andreas Tille wrote:
> Hi Steffen,
> 
> I had a look into MOFA2 and realised you have just issued the needed
> ITPs.  I stumbled upon basikisk which 
> 
>    Installs a self-contained conda instance that is managed by the
>    R/Bioconductor installation machinery
> 
> IMHO this is a no-go since we are simply installing a competing
> package management system via a single package.  I wonder whether
> we can "Freeze Python Dependencies Inside Bioconductor Packages"
> by providing some "fake r-bioc-basilisk" by simply depending from
> the correct Python versions that are needed (and if needed patch
> the consumers ... here MOFA2 accordingly).
> 
> What do you think?

basilisk is also needed for r-bioc-zellkonverter, which is further needed to pass the autopkgtests for
r-bioc-hdf5array

For now, I have disabled tests of hdf5array needing zellkonverter to unbreak the migrations.
But this definitely is a TODO.

I have not gone through the code yet, but if there is a way to disable baslisk and replace the functionality
by something else, that could be a a good way forwd.

But I do wonder if more packages in future might as well end up needing basilisk too.

Cheers,

-- 
Nilesh Patra

Debian Developer, Uploading
  ⢀⣴⠾⠻⢶⣦⠀
  ⣾⠁⢠⠒⠀⣿⡁  
  ⢿⡄⠘⠷⠚⠋   nilesh@debian.org / nileshpatra.info / tchncs.de
  ⠈⠳⣄