romainfrancois

#include <dplyr.h>
// [[Rcpp::depends(BH,dplyr)]]

using namespace Rcpp ;
using namespace dplyr ;

class IndexHelper {
public:
    IndexHelper( int pos_, int n_) : pos(pos_), n(n_){}
    

romainfrancois

Simple R script that grabs the opened issues from a github repo :

$ github_issues hadley/dplyr
   number           user                                                         title                labels
1      98         hadley                                   Setup databases with travis           enhancement
2      97         hadley                                               Unique operator           enhancement
3      96         hadley                           Implement right join and outer join           enhancement
4      95 romainfrancois                                             Hybrid evaluation enhancement, internal
5      94         hadley                 Make sure all vignettes work from R CMD check                   bug

romainfrancois

This uses siesta to pull from travis the log for the last build of a given project.

$ travis_log hadley/dplyr | tail -n30
Le chargement a nécessité le package : methods
Le chargement a nécessité le package : siesta
* checking package vignettes in ‘inst/doc’ ... WARNING
Package vignettes without corresponding PDF/HTML:
   ‘benchmark-baseball.Rmd’
 ‘databases.Rmd’

romainfrancois

require(dplyr)
require(Rcpp)
require(microbenchmark)

# from the cran-logs-dply repo
logs <- readRDS("logs.rds")

# system.time( { by_day <- group_by(logs, date) } )

sourceCpp( "timings.cpp" )

romainfrancois

> f <- function(...) {
+     promises(environment())
+ }

> g <- function(x = 3, ...) {
+     z <- 4
+     f(z = z, ..., x = x)
+ }

> h <- function(..., a = 2) {

romainfrancois

library(Rcpp)
library(microbenchmark)

vectorized <- function() {
  a <- c(1, 1)
  b <- c(2, 2)
  
  for (i in 1:1000000) {
    x <- a + b
  }

romainfrancois

#include <Rcpp.h>
#include <thread>
#include <future>

using namespace Rcpp ;

typedef NumericVector::iterator Iterator ;
  
struct count_positive {
  int operator()(Iterator first, Iterator last){

romainfrancois

ndiff <- 2597797
  f <- function(){
    n <- round(runif( 1, min = 4, max = 10 ))
    paste( sample(letters, n, replace = TRUE ), collapse = "" )
  }
  a <- replicate( ndiff, f() )
  b <- replicate( ndiff, f() )
  
  rows <- sample( 1:ndiff, 3018671, replace = T )
  cols <- sample( c(T,F), 3018671, replace = T )

romainfrancois

> bench(1000)
Unit: microseconds
                     expr    min      lq   median       uq     max neval
                 which(b)  2.104   4.548   5.1000   6.0955  18.306   100
 which_cpp_nthreads(b, 2) 26.401  28.303  28.9285  30.0625 103.980   100
 which_cpp_nthreads(b, 4) 45.902  49.494  50.6335  52.1155 115.184   100
 which_cpp_nthreads(b, 8) 97.596 100.143 101.3720 104.9895 270.521   100

> bench(1e+05)
Unit: microseconds

romainfrancois

library("httr")
library("rjson")

internal <- function(fun){
  names.c <- content( GET("https://raw2.github.com/wch/r-source/trunk/src/main/names.c") )
  names.c <- strsplit( names.c, "\\n" )[[1L]]
  
  rx <- sprintf('^[{]"%s"', fun )
  match <- grep( rx, names.c)
  if( length(match) != 1L ){