Check compatibility with magrittr 2.0

[jashapiro]

Background

Magrittr 2.0 was just released, and while it overall seems great (if soon to be obsolete), it does some changes that could be incompatible with current usage: https://www.tidyverse.org/blog/2020/11/magrittr-2-0-is-here/

Problem

My guess is that we will be fine, but as users who do not use the docker image (most!) may end up with magrittr 2.0, we should check compatibility. The most likely place things might arise is in the timing of evaluation in pipes: magrittr2 is lazy, which can mean that steps that are done for side effects may be skipped.

What potential "gotchas" do we know of?

It's a docker issue, so who knows!

More generally, the "checking if things changed" is always a bit of a challenge, as there will almost certainly be minor changes that need sorting through when doing a diff between html versions.

What are the recommended next steps?

I think the easiest next step is to make an image based on R 4.0.3 (or whatever the now-current version of R is) and check that our pages build with similar results. Using the current R will also result in using the current CRAN and hence will get magrittr2.

If there are any problems that arise, we should aim to fix them in the least disruptive way we can, which will likely be by

Note, I don't think we should unfreeze the refinebio-examples docker image for this; we should stick with the current image of 4.0.2, at least for now. Instead, we should create an alternative dockerfile/image.

Is there a particular timeframe for this issue?

Probably should be done before going live...

A slightly later issue might be to make this a general check we can run to see if we have code that won't work as R, CRAN, *& Bioconductor change and evolve.

[jashapiro]

Oops, wrong reference, reopening.