{bioinformatics,bioconductor}.scm cleaning?

{bioinformatics,bioconductor}.scm cleaning?

[zimoun]

Dear,

Based on the (old) discussion [1] I would like to address this move.

[1] https://lists.gnu.org/archive/html/guix-devel/2018-12/msg00266.html


However, there is already incoherence in bioconductor.scm. :-)


First, the commit c586f427b4 changes `bioconductor-uri` which now
takes optional package type. Thanks Ricardo. :-)
The use of `string-append` in the `uri` field--introduced because of
the annotation and experiment
package--is not expected, right?
Especially, they generally use ".../release/..." without a second
stable URI. Therefore, the future time travel will break.

What is the best: one patch correcting all the packages involved or
one patch per involved package?


Second, all the Bioconductor packages seem to be having a git uri
https://git.bioconductor.org/packages/<name>. And based on the new [2]
`guix lint -c archival` which automatically sends a request to
Software Heritage (SWH) when the fetching method is VCS, it appears to
me a good policy to fix the method to `git-fetch` in the file
`bioconductor.scm`. What do you think ?

From my knowledge, Bioconductor is not yet archived by them. But even
it will be, archive the source in SWH at the "lint time"--i.e., when
the package is modified in Guix--eases their work and ensures
robustness in time travel from our side. From my opinion, replace
`url-fetch` by `git-fetch` will be a good practise.

What do you think?

[2] https://lists.gnu.org/archive/html/guix-patches/2019-08/msg00580.html


Third, bioconductor.scm contains (3) packages which are not in
Bioconductor 3.9; (r-monocle3, r-cicero-monocle3, r-cistopic). Is it
expected?


All the best,
simon