new arrow benchmarks report solution as R-arrow

Author: Tmonster

arrow/groupby-arrow.R renamed to R-arrow/groupby-R-arrow.R

@@ -13,7 +13,7 @@ suppressPackageStartupMessages({
 ver = packageVersion("arrow")
 git = ""
 task = "groupby"
-solution = "arrow"
+solution = "R-arrow"
 fun = "group_by"
 cache = TRUE
 on_disk = FALSE

arrow/join-arrow.R renamed to R-arrow/join-R-arrow.R

@@ -12,7 +12,7 @@ suppressPackageStartupMessages({
 ver = packageVersion("arrow")
 git = ""
 task = "join"
-solution = "arrow"
+solution = "R-arrow"
 cache = TRUE
 on_disk = FALSE
 

R-arrow/setup-R-arrow.sh

@@ -0,0 +1,6 @@
+#!/bin/bash
+set -e
+
+# install stable arrow
+mkdir -p ./arrow/r-arrow
+Rscript -e 'install.packages(c("arrow","dplyr"), lib="./R-arrow/r-arrow")'

arrow/upg-arrow.sh renamed to R-arrow/upg-R-arrow.sh

@@ -3,4 +3,4 @@ set -e
 
 # upgrade all packages in arrow library only if new arrow is out
 echo 'upgrading arrow...'
-Rscript -e 'ap=available.packages(); if (ap["arrow","Version"]!=packageVersion("arrow", lib.loc="./arrow/r-arrow")) update.packages(lib.loc="./arrow/r-arrow", ask=FALSE, checkBuilt=TRUE, quiet=TRUE)'
+Rscript -e 'ap=available.packages(); if (ap["arrow","Version"]!=packageVersion("arrow", lib.loc="./R-arrow/r-arrow")) update.packages(lib.loc="./arrow/r-arrow", ask=FALSE, checkBuilt=TRUE, quiet=TRUE)'

R-arrow/ver-R-arrow.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+set -e
+
+Rscript -e 'v=read.dcf(system.file(package="arrow", lib.loc="./R-arrow/r-arrow", "DESCRIPTION"), fields=c("Version","RemoteSha")); colnames(v)[colnames(v)=="RemoteSha"]="Revision"; cnafill=function(x) {x=c(x); x[is.na(x)]=""; x}; fw=function(f, v) writeLines(v, file.path("arrow", f)); invisible(mapply(fw, toupper(colnames(v)), cnafill(v)))'

_report/report.R

@@ -69,6 +69,9 @@ clean_time = function(d) {
   if (nrow(d[!nzchar(version) | is.na(version)]))
     stop("timings data contains NA or '' as version field, that should not happen")
   old_advanced_groupby_questions = c("median v3 sd v3 by id2 id4","max v1 - min v2 by id2 id4","largest two v3 by id2 id4","regression v1 v2 by id2 id4","sum v3 count by id1:id6")
+
+  # replace arrow with R-arrow (see https://github.com/duckdblabs/db-benchmark/pull/66)
+  d[solution == "arrow", solution := "R-arrow"]
   d[!nzchar(git), git := NA_character_
     ][,"on_disk" := as.logical(on_disk)
       ][task=="groupby" & solution%in%c("pandas","dask","spark") & batch<1558106628, "out_cols" := NA_integer_
@@ -244,8 +247,6 @@ transform = function(ld) {
 
 time_logs = function(path=getwd()) {
   lt <- load_time(path=getwd())
-  # replace arrow with R-arrow (see https://github.com/duckdblabs/db-benchmark/pull/66)
-  lt$solution[lt$solution == "arrow"] <- "R-arrow"
 
   ct = clean_time(lt)
   d = model_time(ct)