[experimental] make use of data.table

[JanMarvin]

A recent post on the r-devel mailinglist reminded me of the data.table package. We have quite a few data frames, or to be precise, I built this package around data frames. There might be some gains using the data.table package, but unfortunately I have not used data.table for quite some time and even when I did, my knowledge was quite limited.

Though I like the package, their long time maintainer and their development ideas. Therefore if we would see some larger gains, I'd be willing to switch to it. Though, I lack time and motivation to do it just for a few microseconds. In the example below (writing a 10.000 x 1.000 matrix, this improves the profvis runtime by about 1.000ms which is nice, but still just 1s.

Since these small performance gains are hard to measure, even benchmarking if something is faster takes a lot of time. And while hunting for some seconds can be fun, more often it is not.

mm <- matrix(1, 1e4, 1e3)

library(openxlsx2)

wb <- wb_workbook()$add_worksheet()

profvis::profvis(wb$add_data(x = mm))

# wb$open()

rm(list = ls())
gc()