mmuurr

Notation

• 0 = red
• 1 = green

The three 'players' are Pa, Pb, and Pc.

Complete global space knowledge (pre-blindfold)

Before any blindfolds are removed, and even before any dots are painted, we have the complete set of possibilities as:

mmuurr

x <- iris
present_names <- c("Sepal.Width", "Sepal.Length")
microbenchmark::microbenchmark(
  dplyr::select(x, present_names),
  dplyr::select(x, dplyr::all_of(present_names)),
  x[present_names],
  x[,present_names]
)

mmuurr

#include <stdio.h>
#include <stdlib.h>
#include <libpq-fe.h>

static void error_exit(PGconn *conn) {
  PQfinish(conn);
  exit(1);
}

int main(int argc, char **argv) {

mmuurr

#include <stdio.h>
#include <stdlib.h>
#include <libpq-fe.h>

static void error_exit(PGconn *conn) {
  PQfinish(conn);
  exit(1);
}

int main(int argc, char **argv) {

mmuurr

#include <stdio.h>
#include <stdlib.h>
#include <libpq-fe.h>

static void error_exit(PGconn *conn) {
  PQfinish(conn);
  exit(1);
}

int main(int argc, char **argv) {

mmuurr

Put test.js in www/ directory for the Shiny app.
Look at console output for different input values.
When input value is 0, the returned value is an object (empty array).
When input value is 1, the returned value is a string.
When input value is >1, the returned value is back to being an object (a populated array).
This forces parsing of returned values that may vary in the length to include conditional logic, which is unneeded complexity.
Proposal: the returned JSON value should conform to jsonlite's default behavior.

mmuurr

## create a boatload of non-trivial functions
my_functions <- lapply(runif(1e6), function(i) function(x) x^i)

system.time(lapply(my_functions, function(f) f(2)))
## output:
##  user  system elapsed
## 1.370   0.023   1.398

x <- list(2)
system.time(lapply(my_functions, do.call, x))