galaaz 2.1.3 → 2.1.5

data/blogs/ruby_plot/ruby_plot.Rmd

@@ -3,7 +3,7 @@ title: "How to make Beautiful Ruby Plots with Galaaz"
 author:
     - "Rodrigo Botafogo"
     - "Daniel Mossé - University of Pittsburgh"
-tags: [Tech, Data Science, Ruby, R, JRuby, "GNU R", Galaaz]
+tags: [Tech, Data Science, Ruby, R, JRuby, CRuby, "GNU R", Galaaz]
 date: "November 19th, 2018 (narrative updated for Galaaz 2.0, 2026)"
 output:
   html_document:
@@ -46,16 +46,16 @@ of libraries for data analysis.
 Until recently, there was no real perspective for Ruby to bridge this gap.
 Implementing a complete scientific computing infrastructure would take too long.
 
-**Galaaz 2.0** couples **[JRuby](https://www.jruby.org/)** (Ruby on the JVM) with **GNU R**—the
-same R distribution used for data science everywhere. A **bridge** evaluates R from Ruby and
-exchanges data between the two processes so that, from Ruby, you call R functions and work
+**Galaaz 2.0** supports **[JRuby](https://www.jruby.org/)** or **CRuby** equally and drives
+**GNU R**—the same R distribution used for data science everywhere. A **bridge** evaluates R from
+Ruby and exchanges data between the two processes so that, from Ruby, you call R functions and work
 with R objects using familiar Ruby syntax. In other words, a Ruby programmer can use the
 capabilities of R without memorizing all of R’s syntax for day-to-day tasks.
 
 An **earlier line of work** used Oracle’s **GraalVM** with **TruffleRuby** and **FastR** so that
 Ruby and R could share one JVM runtime. That stack is **no longer** what Galaaz targets;
-today’s Galaaz is developed and tested with **JRuby + GNU R** (see the project manual for setup
-and command-line tools).
+today’s Galaaz is developed and tested with **JRuby or CRuby + GNU R** (see the project manual for
+setup and command-line tools).
 
 Library wrapping is a common way to bring features from one language into another.
 To improve performance, Python often wraps more efficient C libraries. For the
@@ -84,7 +84,7 @@ ggplot themes that will work the same as the Ruby module.  Yet, writing a new th
 requires specific knowledge on theme writing.  Ruby modules are standard to the
 language and don't need special knowledge.
 
-[Here](https://towardsdatascience.com/ruby-plotting-with-galaaz-an-example-of-tightly-coupling-ruby-and-r-in-graalvm-520b69e21021) is an older article (GraalVM-era Galaaz) with a scatter plot in Ruby; the **ideas** still apply under Galaaz 2.0 with JRuby and GNU R.
+[Here](https://towardsdatascience.com/ruby-plotting-with-galaaz-an-example-of-tightly-coupling-ruby-and-r-in-graalvm-520b69e21021) is an older article (GraalVM-era Galaaz) with a scatter plot in Ruby; the **ideas** still apply under Galaaz 2.0 with JRuby or CRuby and GNU R.
 
 # gKnit
 
@@ -549,12 +549,12 @@ Trying to bring to Ruby the power of R starting from scratch is an enormous ende
 and would probably never be accomplished.  Today's data scientists would certainly
 stick with either Python or R.  Now, both the Ruby and R communities can benefit
 from this marriage: **Galaaz 2.0** uses **standard GNU R** for statistics and graphics
-and **JRuby** for application code, threading, and the JVM ecosystem.  We presented
-the process to couple Ruby and R; the coupling is implemented by the Galaaz bridge and
-**gKnit** for literate documents, not by a single GraalVM polyglot runtime.
+and **JRuby or CRuby** for application code (prefer **JRuby** when you want real threads and the
+JVM ecosystem).  We presented the process to couple Ruby and R; the coupling is implemented by the
+Galaaz bridge and **gKnit** for literate documents, not by a single GraalVM polyglot runtime.
 
 For performance, expect **ordinary GNU R** behaviour for model fitting and plotting, while
-**JRuby** gives **real parallel threads** on the Ruby side and access to Java libraries when
+**JRuby** can give **real parallel threads** on the Ruby side and access to Java libraries when
 you need them.
 
 This article has shown how to improve a plot step-by-step.  Starting from a very simple
@@ -571,8 +571,7 @@ be of great help for any Rubyist trying to write articles, blogs or documentatio
 
 ## Prerequisites (Galaaz 2.0)
 
-* **JRuby** — Ruby on the JVM ([jruby.org](https://www.jruby.org/))
-* A **JDK** compatible with your JRuby version
+* **JRuby** and a compatible **JDK**, *or* **CRuby 3.3+**
 * **GNU R** — `R` on your `PATH`, with compilers/tools available if packages must be built from source
 
 The following R packages will be automatically installed when necessary, but could be installed prior
@@ -594,4 +593,4 @@ are commonly required.
 
 * From the Galaaz repository (or your installed layout), render documents with **`bin/gknit`** (see the project manual for flags such as `--output_format all`).
 * In Ruby code: `require 'galaaz'`
-* For running scripts with the correct JRuby and JVM options, use **`bin/galaaz-jruby`** as described in the manual.
+* Run Ruby with **`bin/galaaz-ruby`** (either engine) or **`bin/galaaz-jruby`** when you need JRuby JVM flags (see the manual).

data/blogs/ruby_plot/ruby_plot.md

@@ -3,7 +3,7 @@ title: "How to make Beautiful Ruby Plots with Galaaz"
 author:
     - "Rodrigo Botafogo"
     - "Daniel Mossé - University of Pittsburgh"
-tags: [Tech, Data Science, Ruby, R, JRuby, "GNU R", Galaaz]
+tags: [Tech, Data Science, Ruby, R, JRuby, CRuby, "GNU R", Galaaz]
 date: "November 19th, 2018 (narrative updated for Galaaz 2.0, 2026)"
 output:
   html_document:
@@ -42,16 +42,16 @@ of libraries for data analysis.
 Until recently, there was no real perspective for Ruby to bridge this gap.
 Implementing a complete scientific computing infrastructure would take too long.
 
-**Galaaz 2.0** couples **[JRuby](https://www.jruby.org/)** (Ruby on the JVM) with **GNU R**—the
-same R distribution used for data science everywhere. A **bridge** evaluates R from Ruby and
-exchanges data between the two processes so that, from Ruby, you call R functions and work
+**Galaaz 2.0** supports **[JRuby](https://www.jruby.org/)** or **CRuby** equally and drives
+**GNU R**—the same R distribution used for data science everywhere. A **bridge** evaluates R from
+Ruby and exchanges data between the two processes so that, from Ruby, you call R functions and work
 with R objects using familiar Ruby syntax. In other words, a Ruby programmer can use the
 capabilities of R without memorizing all of R’s syntax for day-to-day tasks.
 
 An **earlier line of work** used Oracle’s **GraalVM** with **TruffleRuby** and **FastR** so that
 Ruby and R could share one JVM runtime. That stack is **no longer** what Galaaz targets;
-today’s Galaaz is developed and tested with **JRuby + GNU R** (see the project manual for setup
-and command-line tools).
+today’s Galaaz is developed and tested with **JRuby or CRuby + GNU R** (see the project manual for
+setup and command-line tools).
 
 Library wrapping is a usual way of bringing features from one language into another.
 To improve performance, Python often wraps more efficient C libraries. For the
@@ -80,7 +80,7 @@ ggplot themes that will work the same as the Ruby module.  Yet, writing a new th
 requires specific knowledge on theme writing.  Ruby modules are standard to the
 language and don't need special knowledge.
 
-[Here](https://towardsdatascience.com/ruby-plotting-with-galaaz-an-example-of-tightly-coupling-ruby-and-r-in-graalvm-520b69e21021) is an older article (GraalVM-era Galaaz) with a scatter plot in Ruby; the **ideas** still apply under Galaaz 2.0 with JRuby and GNU R.
+[Here](https://towardsdatascience.com/ruby-plotting-with-galaaz-an-example-of-tightly-coupling-ruby-and-r-in-graalvm-520b69e21021) is an older article (GraalVM-era Galaaz) with a scatter plot in Ruby; the **ideas** still apply under Galaaz 2.0 with JRuby or CRuby and GNU R.
 
 # gKnit
 
@@ -622,12 +622,12 @@ Trying to bring to Ruby the power of R starting from scratch is an enormous ende
 and would probably never be accomplished.  Today's data scientists would certainly
 stick with either Python or R.  Now, both the Ruby and R communities can benefit
 from this marriage: **Galaaz 2.0** uses **standard GNU R** for statistics and graphics
-and **JRuby** for application code, threading, and the JVM ecosystem.  We presented
-the process to couple Ruby and R; the coupling is implemented by the Galaaz bridge and
-**gKnit** for literate documents, not by a single GraalVM polyglot runtime.
+and **JRuby or CRuby** for application code (prefer **JRuby** when you want real threads and the
+JVM ecosystem).  We presented the process to couple Ruby and R; the coupling is implemented by the
+Galaaz bridge and **gKnit** for literate documents, not by a single GraalVM polyglot runtime.
 
 For performance, expect **ordinary GNU R** behaviour for model fitting and plotting, while
-**JRuby** gives **real parallel threads** on the Ruby side and access to Java libraries when
+**JRuby** can give **real parallel threads** on the Ruby side and access to Java libraries when
 you need them.
 
 This article has shown how to improve a plot step-by-step.  Starting from a very simple
@@ -644,8 +644,7 @@ be of great help for any Rubyist trying to write articles, blogs or documentatio
 
 ## Prerequisites (Galaaz 2.0)
 
-* **JRuby** — Ruby on the JVM ([jruby.org](https://www.jruby.org/))
-* A **JDK** compatible with your JRuby version
+* **JRuby** and a compatible **JDK**, *or* **CRuby 3.3+**
 * **GNU R** — `R` on your `PATH`, with compilers/tools available if packages must be built from source
 
 The following R packages will be automatically installed when necessary, but could be installed prior
@@ -667,4 +666,4 @@ are commonly required.
 
 * From the Galaaz repository (or your installed layout), render documents with **`bin/gknit`** (see the project manual for flags such as `--output_format all`).
 * In Ruby code: `require 'galaaz'`
-* For running scripts with the correct JRuby and JVM options, use **`bin/galaaz-jruby`** as described in the manual.
+* Run Ruby with **`bin/galaaz-ruby`** (either engine) or **`bin/galaaz-jruby`** when you need JRuby JVM flags (see the manual).

data/examples/misc/ggplot.rb

@@ -151,7 +151,7 @@ R.grid__newpage
 # R.dev__off('')
 # R.awt
 
-(R.ggplot(R.mtcars, E.aes(x: :wt, y: :mpg)) +
+(R.ggplot(~:mtcars, E.aes(x: :wt, y: :mpg)) +
  R.geom_point('')).print
 =end
 

data/lib/R_interface/new_bridge_adapter.rb

@@ -24,7 +24,21 @@ module R
                     File.expand_path('../../ext/new_bridge/galaaz_gatekeeper_phase1.cpp', __dir__)
       @client = NewBridge::SessionClient.new(source_path: source_path)
       @client.start(accept_timeout: 120)
+      bootstrap_session!
+      @ready = true
+    end
 
+    # After a hung/timed-out R eval (e.g. install.packages), kill R and start a fresh runtime
+    # so the compile does not keep running and OOM the host shell.
+    def restart_runtime!(accept_timeout: 120)
+      @ready = false
+      @client.restart!(accept_timeout: accept_timeout)
+      bootstrap_session!
+      @ready = true
+      self
+    end
+
+    def bootstrap_session!
       # Compatibility with the legacy ShadowBridge setup:
       # - `R.awt` / X11 for plotting (examples/sthda_ggplot).
       # - `missing_arg()` for Ruby :all in `[` / tbl subset (R::Support.parse_arg).
@@ -117,7 +131,6 @@ module R
         parent_id: callback_parent_id,
         timeout: init_timeout
       )
-      @ready = true
     end
 
     def ready?

data/lib/R_interface/r.rb

@@ -30,6 +30,7 @@ dir = File.dirname(File.expand_path('.', __FILE__))
 
 # Bridge and Support first
 require_relative 'rsupport'
+require_relative 'r_job'
 
 # Operator modules next (so R::Object can include them)
 require_relative 'rbinary_operators'
@@ -134,68 +135,74 @@ module R
 
   def self.install_rlibs(*libs, install_timeout_sec: nil, callback_timeout_ms: nil, bridge_timeout_sec: nil)
     with_callback_timeout_ms(callback_timeout_ms) do
+      # Await wall-clock for the child Rscript job (nil = wait forever).
+      # Prefer install_timeout_sec / GALAAZ_INSTALL_TIMEOUT_SEC; bridge_timeout_sec
+      # is legacy and only used if the others are unset.
+      await_timeout = install_timeout_sec
+      if await_timeout.nil?
+        raw = ENV['GALAAZ_INSTALL_TIMEOUT_SEC']
+        unless raw.nil? || raw.to_s.strip.empty?
+          await_timeout = begin
+            Integer(raw)
+          rescue StandardError, ArgumentError
+            nil
+          end
+        end
+      end
+      await_timeout = bridge_timeout_sec if await_timeout.nil? && !bridge_timeout_sec.nil?
 
-      # Use a fixed local library directory and ensure R can see it
-      lib_dir = File.expand_path("~/R/x86_64-pc-linux-gnu-library/galaaz")
-      FileUtils.mkdir_p(lib_dir) unless Dir.exist?(lib_dir)
-      R.bridge.eval_r(".libPaths(c('#{lib_dir.gsub("'", "\\\\'")}', .libPaths()))", timeout: bridge_timeout_sec)
+      lib_dir = R::Job.default_lib_dir
+      FileUtils.mkdir_p(lib_dir)
+      R.bridge.eval_r(".libPaths(c('#{lib_dir.gsub("'", "\\\\'")}', .libPaths()))", timeout: 60)
 
       packages = R.c(*libs)
-
-    # installed.packages() returns a matrix; package names are in the row names.
-    # Using [:all, "Package"] on this matrix ends up calling [[ with a missing
-    # row subscript in R, which raises "missing subscript". Instead, rely on
-    # the row names vector for the list of installed packages.
-    installed_mat   = R.installed__packages(nil)
-    installed_names = R.rownames(installed_mat)
-
-    new_packages = packages[!(packages._ :in, installed_names)]
-    new_packages_str = new_packages.to_s
-
-      if(new_packages.length > 0 && new_packages_str != "character(0)")
-      $stderr.puts "[RUBY] The following packages are missing and will be installed: #{new_packages_str}"
-      $stderr.puts "[RUBY] Installing to: #{lib_dir}"
-      
-      # Get packages as Ruby array using :native mode
-      pkg_list = []
-      new_packages.each(:native) { |pkg| pkg_list << pkg }
-      $stderr.puts "[RUBY] Package list: #{pkg_list.inspect}"
-      
-      # Install each package
-      pkg_list.each do |pkg|
-        $stderr.puts "[RUBY] Installing #{pkg}..."
-        if install_timeout_sec
-          timeout_i = Integer(install_timeout_sec)
-          R.bridge.eval_r("options(timeout=#{timeout_i})", timeout: bridge_timeout_sec)
-        end
-        r_cmd = "install.packages('#{pkg.gsub("'", "\\\\'")}', repos='https://cloud.r-project.org', lib='#{lib_dir.gsub("'", "\\\\'")}', dependencies=NA)"
-        $stderr.puts "[RUBY] R command: #{r_cmd}"
-        result = R.bridge.eval_r(r_cmd, timeout: bridge_timeout_sec)
-        $stderr.puts "[RUBY] Install output: #{result.inspect}"
-        if result.to_s.include?("ANTICONF") || result.to_s.include?("Configuration failed")
-          $stderr.puts "[RUBY] WARNING: Package #{pkg} failed to install due to missing system libraries."
-          $stderr.puts "[RUBY] For kableExtra, you may need: libfontconfig1-dev libxml2-dev libfreetype6-dev"
+      installed_mat   = R.installed__packages(nil)
+      installed_names = R.rownames(installed_mat)
+      new_packages = packages[!(packages._ :in, installed_names)]
+      new_packages_str = new_packages.to_s
+
+      if new_packages.length > 0 && new_packages_str != 'character(0)'
+        pkg_list = []
+        new_packages.each(:native) { |pkg| pkg_list << pkg }
+        $stderr.puts "[RUBY] The following packages are missing and will be installed via R::Job: #{pkg_list.inspect}"
+        $stderr.puts "[RUBY] Installing to: #{lib_dir}"
+        $stderr.puts "[RUBY] Bridge R stays free; awaiting child Rscript" +
+                     (await_timeout ? " (timeout=#{await_timeout}s)" : ' (no await timeout)')
+
+        pkg_list.each do |pkg|
+          $stderr.puts "[RUBY] R::Job.install #{pkg}..."
+          begin
+            job = R::Job.install(pkg, lib_dir: lib_dir, wait: true, timeout: await_timeout)
+            job.raise_if_failed!
+            $stderr.puts "[RUBY] R::Job #{job.id} OK for #{pkg}"
+          rescue R::Job::Timeout => e
+            # Child Rscript process group already killed inside Job#wait.
+            raise "R package '#{pkg}' install job timed out. #{e.message}\n" \
+                  "Install outside gknit if needed, then re-run:\n" \
+                  "  Rscript -e \"install.packages('#{pkg}', lib='#{lib_dir}', repos='#{R::Job::DEFAULT_REPOS}')\""
+          rescue R::Job::Failed => e
+            raise "Failed to install R package '#{pkg}' via R::Job.\n#{e.message}"
+          end
         end
-      ensure
-        begin
-          R.bridge.eval_r("options(timeout=60)", timeout: bridge_timeout_sec)
-        rescue StandardError
-          nil
+
+        # Refresh bridge view of libraries after child install.
+        R.bridge.eval_r(".libPaths(c('#{lib_dir.gsub("'", "\\\\'")}', .libPaths()))", timeout: 60)
+        installed_mat = R.installed__packages(nil)
+        installed_names = R.rownames(installed_mat)
+        still_missing = packages[!(packages._ :in, installed_names)]
+        still_missing_str = still_missing.to_s
+        if still_missing.length > 0 && still_missing_str != 'character(0)'
+          raise "Failed to install packages: #{still_missing}. Check ~/.local/share/galaaz/jobs/*/job.log"
         end
       end
-      
-      # Re-check installed packages after install attempt
-      installed_mat = R.installed__packages(nil)
-      installed_names = R.rownames(installed_mat)
-      still_missing = packages[!(packages._ :in, installed_names)]
-      if still_missing.length > 0
-        raise "Failed to install packages: #{still_missing.to_s}. Check stderr output above for [RUBY] debug messages."
-      end
-      end
-
     end
   end
 
+  # Start a CRAN install Job without awaiting (advanced). Prefer +install_and_loads+.
+  def self.install_async(*libs, lib_dir: R::Job.default_lib_dir)
+    R::Job.install(*libs, lib_dir: lib_dir, wait: false)
+  end
+
   #----------------------------------------------------------------------------------------
   #
   #----------------------------------------------------------------------------------------