later2

Schedule an R function or formula to run after a specified period of time. This is similar to JavaScript’s setTimeout function. Like JavaScript, R is single-threaded so there’s no guarantee that the operation will run exactly at the requested time, only that at least that much time will elapse.

To avoid bugs due to reentrancy, by default, scheduled operations only run when there is no other R code present on the execution stack; i.e., when R is sitting at the top-level prompt. You can force past-due operations to run at a time of your choosing by calling later2::run_now().

This package is derived from later, which uses derived work from background package and similar code in Rhttpd.

Installation

You can install the development version of later with:

pak::pak("pachadotdev/later2")

The package is not currently on CRAN.

Usage from R

Pass a function (in this case, delayed by 5 seconds):

later2::later(\() print("Got here!"), 5)

Or a formula (in this case, run as soon as control returns to the top-level):

later2::later(~print("Got here!"))

File Descriptor Readiness

It is also possible to have a function run based on when file descriptors are ready for reading or writing, at an unknown time in the future.

Below, a logical vector is printed indicating which of file descriptors 21 or 22 were ready, subject to a timeout of 1s.

later2::later_fd(\(x) print(x), c(21L, 22L), timeout = 1)

This is useful in particular for asynchronous I/O, allowing reads to be made from TCP sockets as soon as data becomes available. Functions such as curl::multi_fdset() return the relevant file descriptors to be monitored.

Usage from C++

You can also call later2::later from C++ code in your own packages, to cause your own C-style functions to be called back. This is safe to call from either the main R thread or a different thread; in both cases, your callback will be invoked from the main R thread.

later2::later is accessible from later2_api.h and its prototype looks like this:

void later(void (*func)(void*), void* data, double secs)

The first argument is a pointer to a function that takes one void* argument and returns void. The second argument is a void* that will be passed to the function when it’s called back. And the third argument is the number of seconds to wait (at a minimum) before invoking.

later2::later_fd is also accessible from later2_api.h and its prototype looks like this:

void later_fd(void (*func)(int *, void *), void *data, int num_fds, struct pollfd *fds, double secs)

The first argument is a pointer to a function that takes two arguments:

1. An int* array provided by later_fd() when called back, and the second being a void*. The int* array will be the length of num_fds and contain the values 0, 1 or NA_INTEGER to indicate the readiness of each file descriptor, or an error condition respectively.
2. A data object passed to the void* argument of the function when it’s called back. The other required arguments are the total number of file descriptors, a pointer to an array of stuct pollfd, and the number of seconds to wait until timing out.

To use the C++ interface, you’ll need to add later to your DESCRIPTION file under Imports and LinkingTo, and also make sure that your NAMESPACE file has an import(later) entry. See the later2test side package which I created to test later2 to check how to use it from another package.

Background tasks

Finally, this package also offers a higher-level C++ helper class to make it easier to execute tasks on a background thread. It is also available from later2_api.h and its public/protected interface looks like this:

class BackgroundTask {

public:
  BackgroundTask();
  virtual ~BackgroundTask();

  // Start executing the task
  void begin();

protected:
  // The task to be executed on the background thread.
  // Neither the R runtime nor any R data structures may be
  // touched from the background thread; any values that need
  // to be passed into or out of the Execute method must be
  // included as fields on the Task subclass object.
  virtual void execute() = 0;

  // A short task that runs on the main R thread after the
  // background task has completed. It's safe to access the
  // R runtime and R data structures from here.
  virtual void complete() = 0;
}

Create your own subclass, implementing a custom constructor plus the execute and complete methods.

It’s critical that the code in your execute method not mutate any R data structures, call any R code, or cause any R allocations, as it will execute in a background thread where such operations are unsafe. You can, however, perform such operations in the constructor (assuming you perform construction only from the main R thread) and complete method. Pass values between the constructor and methods using fields.

#include "cpp11.hpp"
#include <later2_api.h>

using namespace cpp11;

class MyTask : public later2::BackgroundTask {
public:
  MyTask(doubles vec) :
    inputVals(as_cpp<std::vector<double>>(vec)) {
  }

protected:
  void execute() {
    double sum = 0;
    for (std::vector<double>::const_iterator it = inputVals.begin();
      it != inputVals.end();
      it++) {

      sum += *it;
    }
    result = sum / inputVals.size();
  }

  void complete() {
    Rprintf("Result is %f\n", result);
  }

private:
  std::vector<double> inputVals;
  double result;
};

To run the task, new up your subclass and call begin(), for example (new MyTask(vec))->begin(). There’s no need to keep track of the pointer; the task object will delete itself when the task is complete.

[[cpp11::register]] void asyncMean(doubles data) {
  (new MyTask(data))->begin();
}

Example with parallel jobs (Unix-like systems)

Using the base R ‘parallel’ package, you can fork child processes and obtain real, pollable file descriptors. A child’s fd becomes ready for reading once the child finishes running.

# 1. timeout: prints FALSE, FALSE ----
job1 <- parallel::mcparallel({
  Sys.sleep(1)
  TRUE
})
job2 <- parallel::mcparallel({
  Sys.sleep(1)
  TRUE
})
fd1 <- job1$fd[1]
fd2 <- job2$fd[1]
later_fd(print, c(fd1, fd2), timeout = 0.1)
Sys.sleep(0.2)
run_now()

# 2. fd1 ready: prints TRUE, FALSE ----
job1 <- parallel::mcparallel(TRUE)
fd1 <- job1$fd[1]
Sys.sleep(0.1)
later_fd(print, c(fd1, fd2), timeout = 1)
Sys.sleep(0.1)
run_now()

# 3. both ready: prints TRUE, TRUE ----
job2 <- parallel::mcparallel(TRUE)
fd2 <- job2$fd[1]
Sys.sleep(0.1)
later_fd(print, c(fd1, fd2), timeout = 1)
Sys.sleep(0.1)
run_now()

# 4. fd2 ready: prints FALSE, TRUE ----
parallel::mccollect(job1)
job1 <- parallel::mcparallel({
  Sys.sleep(1)
  TRUE
})
fd1 <- job1$fd[1]
later_fd(print, c(fd1, fd2), timeout = 1)
Sys.sleep(0.1)
run_now()

# 5. fds invalid: prints NA, NA ----
parallel::mccollect(job1)
parallel::mccollect(job2)
later_fd(print, c(fd1, fd2), timeout = 0)
Sys.sleep(0.1)
run_now()

Example with “promise” abstractions

I adapted some functions from the ‘promises’ package using base R. As ‘later’ points out, it is useful to execute tasks on background threads if you cannoy get the results back in R. See the later2test for more “promise” abstractions such as:

# test that resolve works
result <- 0
later2::promise(function(resolve, reject) {
  later2test::asyncFib(resolve, reject, 3)
}) |>
  later2::then(\(x) {
    result <<- x
  })

expect_identical(result, 0)
later2::run_now(1)
while (!later2::loop_empty()) {
  later2::run_now(0.1)
}
expect_identical(result, 2)