Sat, 08 Apr 2017

#4: Simpler shoulders()

Welcome to the fourth post in the repulsively random R ramblings series, or R4 for short.

My twitter feed was buzzing about a nice (and as yet unpublished, ie not-on-CRAN) package thankr by Dirk Schumacher which compiles a list of packages (ordered by maintainer count) for your current session (or installation or ...) with a view towards saying thank you to those whose packages we rely upon. Very nice indeed.

I had a quick look and run it twice ... and had a reaction of ewwww, really? as running it twice gave different results as on the second instance a boatload of tibblyverse packages appeared. Because apparently kids these day can only slice data that has been tidied or something.

So I had another quick look ... and put together an alternative version using just base R (as there was only one subfunction that needed reworking):

source(file="https://raw.githubusercontent.com/dirkschumacher/thankr/master/R/shoulders.R")
format_pkg_df <- function(df) { # non-tibblyverse variant
    tb <- table(df[,2])
    od <- order(tb, decreasing=TRUE)
    ndf <- data.frame(maint=names(tb)[od], npkgs=as.integer(tb[od]))
    colpkgs <- function(m, df) { paste(df[ df$maintainer == m, "pkg_name"], collapse=",") }
    ndf[, "pkg"] <- sapply(ndf$maint, colpkgs, df)
    ndf
}

A nice side benefit is that the function is now free of external dependencies (besides, of course, base R). Running this in the ESS session I had open gives:

R> shoulders()  ## by Dirk Schumacher, with small modifications
                               maint npkgs                                                                 pkg
1 R Core Team <R-core@r-project.org>     9 compiler,graphics,tools,utils,grDevices,stats,datasets,methods,base
2 Dirk Eddelbuettel <edd@debian.org>     4                                  RcppTOML,Rcpp,RApiDatetime,anytime
3  Matt Dowle <mattjdowle@gmail.com>     1                                                          data.table
R> 

and for good measure a screenshot is below:

I think we need a catchy moniker for R work using good old base R. SoberVerse? GrumbyOldFolksR? PlainOldR? Better suggestions welcome.

Edit on 2017-04-09: And by now Dirk Schumacher fixed that little bug in thankr which was at the start of this. His shoulders() function is now free of side effects, and thankr is now a clean micropackage free of external depends from any verse, be it tiddly or grumpy. I look forward to seeing it on CRAN soon!

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

/code/r4 | permanent link