Re: Rcwl

[zimoun <zimon.toutoune@gmail.com>]

Hi Pjotr,


On Thu, 21 Feb 2019 at 17:56, Pjotr Prins <pjotr.public12@thebird.nl> wrote:
>
> This is a recent implementation of CWL in R:
>
>   https://github.com/hubentu/Rcwl

If I understand well, it is "just" a R front-end to cwltool.
Basically, it reads CWL files, it writes CWL files from R flavored
pipeline description and it runs CWL tools.

Naive question: what does they gain?
CWL extended by R?


From my opinion, the approach by SnakeMake is more interesting.
 1- they describes the workflow with snakemake and so it is natively
extendable with Python; then the "graph" is exported to CWL file.
 2- they imports a complete CWL workflow as a snakemake rule; then
this rule runs cwltool, and can be connected to any other snakemake
rules.

[1] https://snakemake.readthedocs.io/en/stable/executable.html#cwl-export
[2] https://snakemake.readthedocs.io/en/stable/snakefiles/modularization.html#common-workflow-language-cwl-support


> I am thinking for future use of GWL it would be of interest to support
> CWL output and maybe even a CWL runner. Maybe a Guile project for
> someone who wants to have a go.

We already had a discussion about this idea on guix-devel, one year ago. :-)

https://lists.gnu.org/archive/html/guix-devel/2018-01/msg00428.html

I agree that a "CWL engine" would be nice. I mean my mind has not
changed so much since last year. ;-)


What is blocking to push your guix-cwl channel to Guix master?


All the best
--
simon