[general] Does this package lock users into R code only? if so are there any plans so the package can call other executables

[oTheAnalyst]

Help

• I understand and agree to https://books.ropensci.org/targets/help.html.

Description

Are there any plans to add a backend interface so you could run the pipelines on various cluster backends like SLURM, AWS Batch, etc?

[Aariq]

You can run targets pipelines on various backends with the crew package and crew.cluster package including SLURM and AWS. There are some examples in the manual.

As for running things besides R code, you can create R code wrappers for calling commandline tools with system2() or the processx package, or you can run python code from R with reticulate. I think creating these kinds of wrappers in R is usually your best be for integrating other things besides R with targets

[kendonB]

I have had some python-based junior colleagues try out snakemake after I had such a good experience with targets. They didn't seem to easily adopt snakemake like my R-based colleagues had been able to pick up targets (and drake before that). If anyone has an opinion, would it be easier / better to try to use targets rather than using snakemake for someone who mostly uses python?

[ginolhac]

Snakemake is useful and for bioinformatics a must have.
But building dependencies from the bottom (rule all) and wildcards are making writing of new pipelines much more difficult than in targets.
Also it is lacking the debugging facilities of targets.
For users, good templates are very nice to have and use. I guess this is the core aspect of snakemake.
Targets is more about writing your analysis with clear dependencies than applying recipes.