Package future

[seabbs]

I think as the pandemic progresses more and more data sets will stop being supported. There are quite a few options about what we could do with the package to deal with this.

One potential option is that we split the package into multiple packages. One that has the tools to download, process, clean, and test data and one or several that contain the supported datasets. This would make long term support much easier as well as freeing up using the tools for future non-COVID-19 related projects.

Another option is to keep maintaining as we are with a third option being some kind of retirement once a given number of datasets are no longer supported.

Opening this discussion as a place to start thinking about how/if/when to make these or other changes.

[seabbs]

@epiforecasts/covidregionaldata @RichardMN @Bisaloo @sbfnk

[RichardMN]

Comments, a bit random...

I don't have a clear sense of how broadly the package is (still) being used. I am using it for personal interest/tracking/comparison purposes.

I think the modular approach to dataset provision is an interesting idea. It may need some thinking as to how the dataset modules are developed/published/installed - I don't think we're going to have a CRAN entry for each Class but we could probably get to the point where a dataset module is a single downloadable R file which a savvy user can install using a package-provided command. (I'm not sure I'd suggest going to another package repository outside CRAN if there's a way to do it with people just grabbing a file from github or choosing it from a local directory.)

The modular approach would break out the not-so-interesting grind of fixing dataset glitches. I'd nonetheless welcome having some form of peer/code review available for trying to fix those datasets. As seen getting to 0.9.3 I've made some mistakes or misunderstood things and it's really useful having another set of (better-qualified) eyes look at the code.

I have learned a lot working on this package with the epiforecasts team and enjoy it. This is an evenings/weekends activity for me. It's up to the main developers/users to decide if it's worth the time and effort to keep going, or if there's a graceful senescence approach available.