Finding packages

R, as I’ve pointed out before, has a package discovery problem.

There’s a new package, colorblindr, which lets you see the impact of various sorts of colour-blindness on a colour palette, a very useful thing for designing good graphics. When it’s mentioned on Twitter, you see lots of people glad that such a tool is now available for R.

A similar tool has been available on CRAN for fifteen years.

Now, the new package looks to have a better representation of colour vision. The dichromat package used colour-matching experiments from the 1990s and interpolates with loess; colorblindr seems to have a unified colour model.   It would be nice to see a comparison of the two packages, and maybe it would show everyone should move to the new one.  That’s what’s happened to a lot of my contributions to R. Progress is good.

But my point right now is there are a lot of people excited about colorblindr who didn’t know dichromat existed. Most of these are people who would have been interested in dichromat but didn’t know about it.  I haven’t exactly been quiet about the package –for example, it’s in the R courses that Ken Rice and I teach around the world, and I wrote about it for the ASA Computing and Graphics newletter.  It’s not hard to get me to talk about color spaces; some people would say it’s hard to get me to stop.

CRAN Task Views have helped a bit with the package discovery problem – but dichromat is in the Graphics task view. It helps if a popular package suggests yours – but dichromat is a dependency of a dependency of ggplot2. It helps if the package author is reasonably well known or the package was on CRAN early – but again. It would help to give a clearer name, perhaps?

There must be lots of people out there who would be excited to find there was a package doing just what they needed. But there isn’t an easy way for them to notice.

Original Source.

Feature Gif Source.