Get USAboundaries and USAboundariesData on CRAN about usaboundaries

[hadley]

How much of a difference would it make if you split the data package into historical and modern datasets?

[lmullen]

It might be worth doing. The historical data is only about 2.4 Mb. But I'd like to add higher resolution versions, which will add quite a bit. (If I do that, then the resolution argument will change from resolution = c("20m", "5m", "500k") to resolution = c("low", "med", "high"). But the contemporary data will have to change every year, whereas the historical will never change.

The single biggest file is the zip codes, which is 32 Mb.

[hadley]

Hmmmmm, maybe separate out the zipcodes? CRAN is definitely not going to go for a 30 meg package that gets updated every year.

Bigger data packages can go as drat() repos - and you can include notes describing how to install, or provide a helper function that calls install.packages() with the right arguments. Does that make sense?

[lmullen]

Thanks for the suggestion, @hadley. I have the same basic problem with the gender package. But it looks like rOpenSci has a drat repository up and running. I think I might try that approach first for the data only packages, which will give me more flexibility about adding and extending, then just put the interface package on CRAN.

[hadley]

I'd recommend putting at least one data package on cran - then the package can do something straight away. And have examples etc.

[lmullen]

This issue is fixed in the latest release. I included the contemporary and historical low-res data in the CRAN package, which can install a package with higher-resolution data from the rOpenSci repository.

[hadley]

Perfect 😄